{ "amendments": [ { "id": "A-L01", "new": "swap used < 80 % AND MemAvailable >= 10 GiB AND PSI memory full avg60 < 10 % (/dev/shm < 4 GiB kept)", "old": "swap used < 50 %, /dev/shm used < 4 GiB, MemAvailable >= 10 GiB", "readings_at_1824Z": { "MemAvailable_GiB": 49.2, "MemFree_GiB": 41.0, "bench_unit_MemorySwapCurrent_bytes": 2994176, "largest_swap_holders": "two stale sharpsignal.memory_mcp processes (pids 9106, 12122; ~1.1 GB each) in old ghostty scopes, and the system ollama's nomic runner (0.34 GB)", "psi_memory_full_avg60_pct": 1.74, "swap_img": "0.3 of 2.0 GiB", "zram0": "9.9 of 16.0 GiB used, stored in 3.56 GiB of RAM" }, "reason": "the old gate tripped at 18:24:03Z (dry12-anchor-4096/REFUSED.json: swap 54.2 %) with no memory risk: this box's swap is mostly zram (compressed RAM), so a bare percentage over-reads the risk; the new gate still catches the 15:56Z case (swap 17 of 17 GiB)", "rows_the_author_had_seen": "none scored: the gate tripped before item 12's first load", "ruled_by": "the lead under R-0 (message to the laptop lane, received ~18:25Z; applied 18:25:59Z)", "utc": "2026-09-23T18:25Z", "what": "the memory gate (PLAN-v2 §7, G-2)" }, { "id": "A-L02", "new": "MemoryMax=16G", "old": "MemoryMax=4G", "reason": "item 14 measured: under 4G the serve unit hit its cap 633,419 times (memory.events max) during item 12, swapped up to 6.0 GB (MemorySwapPeak 6,008,573,952 B), memory.pressure full avg60 read 9.41 %, and prefill stalled 11,416 ms on doorman item A5 and 4,607 ms on C1 (515-521 prompt tokens; the other items 121-342 ms). No OOM, but the cap throttled the measurement. 16G is a ceiling that does not bind: llama-server's host prompt cache is bounded (~8 GiB) and the rest is clean page cache; each arm.json records memory.events, memory.stat and memory.pressure before and after, so a binding cap is visible", "rows_the_author_had_seen": "dry12-anchor-4096 and dry12-anchor-131072 under 4G (kept as superseded receipts)", "ruled_by": "the lane, under PLAN-v2 Appendix A/B item 14 ('raise to a measured value if it OOMs; never remove')", "utc": "2026-09-23T18:31Z", "what": "the bench instance's MemoryMax (Appendix A: 4G, marked UNTESTED)" }, { "id": "A-L03", "new": "MemoryMax=32G (applied live: systemctl --user set-property)", "old": "MemoryMax=16G (A-L02)", "reason": "16G still bound: llama-server's host prompt cache grows to its own 8,192 MiB limit (its log: '54 prompts, 8139.974 MiB (limits: 8192.000 MiB ...)'), and with an ~18 GB model file's page cache charged to the same cgroup the cap was hit 5,598 times during k0a-gemma4-anchor-cd and 6,994 times during b-vicuna-33b; both arms show 1.4-7.2 s prefill stalls on 1-3 items that benchbox's same arms do not. Anon stays <= ~9 GB; the rest is reclaimable page cache, so 32G is a ceiling that should not bind, and every arm.json carries the cgroup's memory.events before/after to prove it", "rows_the_author_had_seen": "k0a-gemma4-anchor-cd, k1-mistral-v0.1-cd, b-falcon-7b, b-vicuna-33b, and b-llama2-7b's task c", "ruled_by": "the lane, under PLAN-v2 Appendix B item 14 (as A-L02)", "utc": "2026-09-23T18:51:52Z", "what": "the bench instance's MemoryMax" }, { "id": "A-L04", "new": "nvidia-powerd active, clocks.max.sm 3090 (lock off), AC online, one board, and enforced.power.limit anywhere in the Dynamic Boost envelope (> 95 W and <= 175 W); the limit is reported per task as min/median/max from the 1 Hz ext trace, never as one number", "old": "enforced.power.limit == 150.00 W (+-0.5), clocks.max.sm 3090, powerd active, AC on, one board", "reason": "under Dynamic Boost the firmware moves the GPU's enforced limit when the CPU draws: the 1 Hz traces read 135.00-150.00 W on b-falcon-7b and b-vicuna-33b and 137.08-150.00 W on b-llama2-7b (a steady 150.00 on the anchor and mistral; loadavg ~5 from peers); the old gate refused b-llama2-7b before its task d at 18:51:55Z (146.76 W). Memory laptop-5090-refuses-a-power-cap: the boost posture's cap is 'a range from the trace'", "rows_the_author_had_seen": "the same as A-L03", "ruled_by": "the lead under R-0 (message to the laptop lane, received 18:53Z)", "utc": "2026-09-23T18:53Z", "what": "the laptop posture gate (PLAN-v2 §2.2: 'enforced.power.limit 150.00 W ... gated per arm, rc 4 on mismatch')" } ], "argv": [ "coilaptop.py", "--arm-id", "b-qwen3.5-27b", "--out", "/workshop/estate/bench/cost-of-intelligence-2026-09-23/laptop-reading-0/rows/b-qwen3.5-27b", "--model", "qwen3.5:27b-q4_K_M", "--num-ctx", "4096", "--think", "off", "--think-budget", "4096", "--tasks", "c,d", "--probes" ], "arm_id": "b-qwen3.5-27b", "baseline_rows": [ { "memory_mib": 15.0, "sm_mhz": 1192.0, "temp_c": 39.0, "uuid": "", "watts": 11.27 } ], "board": "", "box": "the laptop", "box_class": "gpu-5090-laptop-24g", "digest_pin": "7653528ba5cb", "disk_floor": { "df": "Avail\n113722642432", "floor_gib": 60.0, "free_gib": 105.9124641418457, "store": "/workshop/bench-coi-2026-09-23/store" }, "ended": "2026-09-23T21:23:12Z", "energy_instruments": [ "board (nvidia-smi power.draw by UUID, 1 Hz; run.py Boards)", "rapl psys + package-0 (+ core, uncore) via /run/rapl/energy (R-10d); not wall" ], "host": "http://127.0.0.1:withheld-2026-10-14", "idle_preload": { "board": { "cards": [ "" ], "joules": 168.95689462065698, "mean_watts": 12.062318482813573, "peak_watts": 19.24, "per_card": { "": { "joules": 168.95689462065698, "mean_watts": 12.062318482813573, "min_watts": 11.15, "peak_watts": 19.24, "samples": 15, "seconds": 14.006999969482422 } }, "samples": 15, "seconds": 14.006999969482422 }, "rapl": { "domains": { "core": { "joules": 582.99271, "mean_watts": 43.529514264539145, "samples": 14, "seconds": 13.393044233322144, "wraps": 0 }, "package-0": { "joules": 702.7753, "mean_watts": 52.47483469624089, "samples": 14, "seconds": 13.392615795135498, "wraps": 0 }, "psys": { "joules": 3.620291, "mean_watts": 0.27025953735484515, "samples": 14, "seconds": 13.39560866355896, "wraps": 0 }, "uncore": { "joules": 0.363463, "mean_watts": 0.027135646802379936, "samples": 14, "seconds": 13.39430022239685, "wraps": 0 } }, "instrument": "RAPL via /run/rapl/energy (root sampler, 1 Hz; R-10d): psys = the platform's own reading (not wall); package-0 = the CPU package" }, "window_epoch": [ 1790198343.897, 1790198357.904 ] }, "instrument": { "coilaptop_py": "0b5dcb41ac989a57dfe4912740dfd24ccd600da5ba3fbc181688d18a8adf94f0", "jev/kit/doorman_planted.json": "4daf059c26a47a84fa3b73630a64ba1a348da72c2e5df7f3c28c007b67da22f7", "jev/kit/task_c.json": "61a0b1d6e7cf66fd91992c0c3a4572f6fbae90e6471048fd7dc1eb41ff2d7f39", "jev/run.py": "withheld-2026-10-14", "jev/run_addenda.py": "a34aca61833ed0c5860c91c7e2e227429e8ea9e2e50fec68d7b7fec15d89acab" }, "limit": null, "memory_gate": { "amendment": "A-L01", "limits": { "mem_available_gib_min": 10.0, "psi_full_avg60_max_pct": 10.0, "shm_used_gib_max": 4.0, "swap_used_frac_max": 0.8 }, "mem_available_gib": 41.861270904541016, "psi_memory": { "full": { "avg10": 0.95, "avg300": 2.78, "avg60": 4.6, "total": 118476038.0 }, "some": { "avg10": 0.97, "avg300": 2.95, "avg60": 4.93, "total": 126366135.0 } }, "shm_used_gib": 3.8918228149414062, "stamp": "2026-09-23T21:18:58Z", "swap_total_gib": 17.99999237060547, "swap_used_frac": 0.3433, "swaps": [ "Filename\t\t\t\tType\t\tSize\t\tUsed\t\tPriority", "/swap.img file\t\t2097148\t\t59120\t\t-1", "/dev/zram0 partition\t16777212\t6420648\t\t100" ], "why": "ok" }, "model": "qwen3.5:27b-q4_K_M", "num_ctx": 4096, "ollama_version": "0.32.13", "pin": { "api_ps_model": { "context_length": 4096, "details": { "families": [ "qwen35" ], "family": "qwen35", "format": "gguf", "parameter_size": "27.8B", "parent_model": "", "quantization_level": "Q4_K_M" }, "digest": "7653528ba5cba4dd8e19da24aaddc7f4d0b5ecd93571c0825dfd4137958ec06e", "expires_at": "2319-01-03T21:07:23.056053673Z", "model": "qwen3.5:27b-q4_K_M", "name": "qwen3.5:27b-q4_K_M", "size": 16325279742, "size_vram": 16325279742 }, "card": "", "cards": [ "" ], "fit": { "growth_bytes": 17720934400.0, "ps_size": 16325279742, "ps_size_vram": 16325279742, "ps_whole": true, "rule": "WHOLE when the named board grew by at least 90 %% of the store's bytes for this tag (weights + projector/draft; KV rides on top) AND /api/ps does not report a CPU share; ps alone is corroboration, never the verdict", "store_bytes": 17420432728, "whole": true }, "floor_mib": 200.0, "growth_mib": { "": 16900.0 }, "memory_after_mib": { "": 16915.0 }, "named_grew": true, "ok": true, "others_quiet": true, "per_card_grew": { "": true } }, "postflight": { "after_unload": { "api_ps": { "models": [] }, "api_ps_empty": true, "back_to_baseline": true, "board_mib_baseline": 15.0, "board_mib_now": 15.0, "rows": [ { "memory_mib": 15.0, "sm_mhz": 1192.0, "temp_c": 46.0, "uuid": "", "watts": 16.46 } ], "stamp": "2026-09-23T21:23:12Z" }, "posture_after": { "ac_online": "1", "battery": { "capacity": "93", "status": "Not charging" }, "boards_on_bus": [ "0, , NVIDIA GeForce RTX 5090 Laptop GPU, withheld-2026-10-14" ], "card": { "clocks.max.sm": "3090", "clocks.mem": "14001", "clocks.sm": "1965", "clocks_throttle_reasons.active": "0x0000000000000000", "driver_version": "595.91.07", "enforced.power.limit": "150.00", "memory.total": "24463", "memory.used": "16927", "name": "NVIDIA GeForce RTX 5090 Laptop GPU", "pci.bus_id": "withheld-2026-10-14", "pcie.link.gen.current": "5", "pcie.link.gen.max": "5", "pcie.link.width.current": "8", "pcie.link.width.max": "16", "persistence_mode": "Enabled", "power.default_limit": "95.00", "power.draw": "95.94", "power.limit": "[N/A]", "power.max_limit": "175.00", "pstate": "P0", "temperature.gpu": "55", "utilization.gpu": "9", "uuid": "", "vbios_version": "98.03.A2.00.21" }, "compute_apps": [ ", 55051, /workshop/bench-coi-2026-09-23/ollama-0.32.13/lib/ollama/llama-server, 16904 MiB" ], "fan": "unreadable on this board (fan.speed N/A); the operator's turbo fan per OPERATOR-CONDITIONS.md", "how": "nvidia-smi -i ; enforced.power.limit and clocks.max.sm READ, never typed; power.limit reads N/A on this board and the board refuses -pl", "loadavg": "2.67 3.01 2.82 1/1910 101424", "nvidia_powerd": "active", "stamp": "2026-09-23T21:23:06Z" }, "seat_after": { "api_ps": { "models": [] }, "stamp": "2026-09-23T21:23:12Z" }, "serve_cgroup_memory_after": { "events": { "high": "0", "low": "0", "max": "43078", "oom": "0", "oom_group_kill": "0", "oom_kill": "0", "sock_throttled": "401" }, "pressure": [ "some avg10=0.00 avg60=0.00 avg300=0.02 total=15976543", "full avg10=0.00 avg60=0.00 avg300=0.02 total=15976543" ], "stat": { "anon": 51949568, "file": 2374119424, "file_mapped": 1757184, "shmem": 0 }, "swap_current": 2633728 }, "serve_unit_after": { "ActiveState": "active", "ControlGroup": "/user.slice/user-1000.slice/user@1000.service/app.slice/coi-ollama-laptop.service", "Environment": "TZ=UTC OLLAMA_HOST=127.0.0.1:withheld-2026-10-14 OLLAMA_KEEP_ALIVE=-1 OLLAMA_MAX_LOADED_MODELS=1 OLLAMA_=1 OLLAMA_FLASH_ATTENTION=1 OLLAMA_KV_CACHE_TYPE=f16 OLLAMA_NOPRUNE=1 OLLAMA_MODELS=/workshop/bench-coi-2026-09-23/store CUDA_VISIBLE_DEVICES= OLLAMA_VULKAN=0 GGML_VK_VISIBLE_DEVICES=-1 HIP_VISIBLE_DEVICES= ROCR_VISIBLE_DEVICES=", "ExecStart": "{ path=/workshop/bench-coi-2026-09-23/ollama-0.32.13/bin/ollama ; argv[]=/workshop/bench-coi-2026-09-23/ollama-0.32.13/bin/ollama serve ; ignore_errors=no ; start_time=[n/a] ; stop_time=[n/a] ; pid=0 ; code=(null) ; status=0/0 }", "MainPID": "1739347", "MemoryCurrent": "2447900672", "MemoryMax": "34359738368", "MemoryPeak": "30671613952", "NRestarts": "0" } }, "posture_before": { "ac_online": "1", "battery": { "capacity": "93", "status": "Not charging" }, "boards_on_bus": [ "0, , NVIDIA GeForce RTX 5090 Laptop GPU, withheld-2026-10-14" ], "card": { "clocks.max.sm": "3090", "clocks.mem": "810", "clocks.sm": "1192", "clocks_throttle_reasons.active": "0x0000000000000000", "driver_version": "595.91.07", "enforced.power.limit": "150.00", "memory.total": "24463", "memory.used": "15", "name": "NVIDIA GeForce RTX 5090 Laptop GPU", "pci.bus_id": "withheld-2026-10-14", "pcie.link.gen.current": "2", "pcie.link.gen.max": "5", "pcie.link.width.current": "8", "pcie.link.width.max": "16", "persistence_mode": "Enabled", "power.default_limit": "95.00", "power.draw": "11.83", "power.limit": "[N/A]", "power.max_limit": "175.00", "pstate": "P5", "temperature.gpu": "39", "utilization.gpu": "0", "uuid": "", "vbios_version": "98.03.A2.00.21" }, "compute_apps": [], "fan": "unreadable on this board (fan.speed N/A); the operator's turbo fan per OPERATOR-CONDITIONS.md", "how": "nvidia-smi -i ; enforced.power.limit and clocks.max.sm READ, never typed; power.limit reads N/A on this board and the board refuses -pl", "loadavg": "6.47 3.43 2.83 4/1897 45300", "nvidia_powerd": "active", "stamp": "2026-09-23T21:18:58Z" }, "posture_verdict": "Dynamic Boost: enforced 150.00 W (inside the (95, 175] W envelope), clock lock off (3090 MHz), powerd active, AC on, one board", "probes": { "decode": { "board": { "cards": [ "" ], "joules": 2907.6997543287275, "mean_watts": 126.36129523672915, "peak_watts": 138.36, "per_card": { "": { "joules": 2907.6997543287275, "mean_watts": 126.36129523672915, "min_watts": 37.24, "peak_watts": 138.36, "samples": 24, "seconds": 23.01099991798401 } }, "samples": 24, "seconds": 23.01099991798401 }, "num_predict": 256, "prompt": "Describe, in about three hundred words, a lighthouse on a rocky coast during a winter storm.", "rapl": { "domains": { "core": { "joules": 515.203393, "mean_watts": 22.697792404505538, "samples": 23, "seconds": 22.698392152786255, "wraps": 0 }, "package-0": { "joules": 687.603708, "mean_watts": 30.293063639722217, "samples": 23, "seconds": 22.69838786125183, "wraps": 0 }, "psys": { "joules": 6.107712, "mean_watts": 0.2690900378963835, "samples": 23, "seconds": 22.697651863098145, "wraps": 0 }, "uncore": { "joules": 0.642394, "mean_watts": 0.02830168411858727, "samples": 23, "seconds": 22.698083877563477, "wraps": 0 } }, "instrument": "RAPL via /run/rapl/energy (root sampler, 1 Hz; R-10d): psys = the platform's own reading (not wall); package-0 = the CPU package" }, "runs": [ { "decode_tok_s": 37.100395409057946, "done_reason": "length", "eval_count": 256, "eval_s": 6.900196, "load_s": 0.157116478, "prefill_tok_s": 158.97698311366358, "prompt_eval_count": 32, "prompt_eval_s": 0.201287, "runner_pids": [ 55051, 1739347 ], "text_head": "The lighthouse stands as a solitary sentinel against the churning fury of the winter sea, perched precariously upon jagg", "thinking_chars": 0, "total_s": 7.517588666 }, { "decode_tok_s": 37.11634960701615, "done_reason": "length", "eval_count": 256, "eval_s": 6.89723, "load_s": 0.180844684, "prefill_tok_s": 276.4650487701625, "prompt_eval_count": 32, "prompt_eval_s": 0.115747, "runner_pids": [ 55051, 1739347 ], "text_head": "The lighthouse stands as a solitary sentinel against the churning fury of the winter sea, perched precariously upon jagg", "thinking_chars": 0, "total_s": 7.408763706 }, { "decode_tok_s": 37.08573889640664, "done_reason": "length", "eval_count": 256, "eval_s": 6.902923, "load_s": 0.209469666, "prefill_tok_s": 291.8243582143997, "prompt_eval_count": 32, "prompt_eval_s": 0.109655, "runner_pids": [ 55051, 1739347 ], "text_head": "The lighthouse stands as a solitary sentinel against the churning fury of the winter sea, perched precariously upon jagg", "thinking_chars": 0, "total_s": 7.225156964 } ], "window_epoch": [ 1790198539.051, 1790198562.062 ] }, "prefill": { "board": { "cards": [ "" ], "joules": 1527.3986366117, "mean_watts": 117.40189375336908, "peak_watts": 152.42, "per_card": { "": { "joules": 1527.3986366117, "mean_watts": 117.40189375336908, "min_watts": 35.87, "peak_watts": 152.42, "samples": 14, "seconds": 13.009999990463257 } }, "samples": 14, "seconds": 13.009999990463257 }, "rapl": { "domains": { "core": { "joules": 239.292662, "mean_watts": 21.08390267385433, "samples": 12, "seconds": 11.34954309463501, "wraps": 0 }, "package-0": { "joules": 327.90242, "mean_watts": 28.89155727419231, "samples": 12, "seconds": 11.349420070648193, "wraps": 0 }, "psys": { "joules": 3.084403, "mean_watts": 0.2717765719256405, "samples": 12, "seconds": 11.349039316177368, "wraps": 0 }, "uncore": { "joules": 0.288817, "mean_watts": 0.025447458268616358, "samples": 12, "seconds": 11.349542140960693, "wraps": 0 } }, "instrument": "RAPL via /run/rapl/energy (root sampler, 1 Hz; R-10d): psys = the platform's own reading (not wall); package-0 = the CPU package" }, "runs": [ { "decode_tok_s": 1000000.0, "done_reason": "length", "eval_count": 1, "eval_s": 1e-06, "load_s": 0.177853713, "prefill_tok_s": 1433.3798247514228, "prompt_eval_count": 2263, "prompt_eval_s": 1.578786, "runner_pids": [ 55051, 1739347 ], "text_head": "Ev", "thinking_chars": 0, "total_s": 1.770802965 }, { "decode_tok_s": 1000000.0, "done_reason": "length", "eval_count": 1, "eval_s": 1e-06, "load_s": 0.202127891, "prefill_tok_s": 1430.324653963916, "prompt_eval_count": 2261, "prompt_eval_s": 1.58076, "runner_pids": [ 55051, 1739347 ], "text_head": "Ev", "thinking_chars": 0, "total_s": 2.102223526 }, { "decode_tok_s": 1000000.0, "done_reason": "length", "eval_count": 1, "eval_s": 1e-06, "load_s": 0.179740633, "prefill_tok_s": 1431.3792243721737, "prompt_eval_count": 2262, "prompt_eval_s": 1.580294, "runner_pids": [ 55051, 1739347 ], "text_head": "Ev", "thinking_chars": 0, "total_s": 2.057417197 }, { "decode_tok_s": 1000000.0, "done_reason": "length", "eval_count": 1, "eval_s": 1e-06, "load_s": 0.186024585, "prefill_tok_s": 1428.3205452002492, "prompt_eval_count": 2261, "prompt_eval_s": 1.582978, "runner_pids": [ 55051, 1739347 ], "text_head": "Ev", "thinking_chars": 0, "total_s": 2.074180397 }, { "decode_tok_s": 1000000.0, "done_reason": "length", "eval_count": 1, "eval_s": 1e-06, "load_s": 0.194759921, "prefill_tok_s": 1429.9736111760271, "prompt_eval_count": 2264, "prompt_eval_s": 1.583246, "runner_pids": [ 55051, 1739347 ], "text_head": "Ev", "thinking_chars": 0, "total_s": 2.071626675 }, { "decode_tok_s": 1000000.0, "done_reason": "length", "eval_count": 1, "eval_s": 1e-06, "load_s": 0.206464583, "prefill_tok_s": 1424.1035995967877, "prompt_eval_count": 2259, "prompt_eval_s": 1.586261, "runner_pids": [ 55051, 1739347 ], "text_head": "Ev", "thinking_chars": 0, "total_s": 2.082460685 } ], "target_tokens_approx": 3000.0, "window_epoch": [ 1790198563.104, 1790198576.114 ] }, "ttft": { "how": "the first 12 task_c prompts, run.py's body with stream true, in task order (so the prompt-cache condition is the task's own); TTFT = client clock on localhost from the request to the first chunk carrying content or thinking", "median_s": 0.7585990484949434, "p95_s": 0.8087417390051996, "runs": [ { "eval_count": 2, "prompt_eval_count": 228, "server_load_ms": 183.988297, "server_prompt_eval_ms": 279.922, "total_s": 0.8102349570108345, "ttft_s": 0.7817581679992145 }, { "eval_count": 2, "prompt_eval_count": 324, "server_load_ms": 177.827322, "server_prompt_eval_ms": 343.763, "total_s": 0.7594284739898285, "ttft_s": 0.7306228160014143 }, { "eval_count": 2, "prompt_eval_count": 245, "server_load_ms": 188.838339, "server_prompt_eval_ms": 288.796, "total_s": 0.7360011249984382, "ttft_s": 0.7079348639963428 }, { "eval_count": 3, "prompt_eval_count": 346, "server_load_ms": 196.211936, "server_prompt_eval_ms": 357.224, "total_s": 0.8412877890077652, "ttft_s": 0.7841049220005516 }, { "eval_count": 3, "prompt_eval_count": 356, "server_load_ms": 205.771376, "server_prompt_eval_ms": 365.928, "total_s": 0.8642023200081894, "ttft_s": 0.8087417390051996 }, { "eval_count": 2, "prompt_eval_count": 358, "server_load_ms": 168.778136, "server_prompt_eval_ms": 355.501, "total_s": 0.7890387020015623, "ttft_s": 0.7614738449919969 }, { "eval_count": 2, "prompt_eval_count": 263, "server_load_ms": 174.930535, "server_prompt_eval_ms": 309.137, "total_s": 0.7415168720035581, "ttft_s": 0.7134009779983899 }, { "eval_count": 2, "prompt_eval_count": 244, "server_load_ms": 190.903757, "server_prompt_eval_ms": 294.315, "total_s": 0.7546881909947842, "ttft_s": 0.726097648992436 }, { "eval_count": 2, "prompt_eval_count": 320, "server_load_ms": 178.281272, "server_prompt_eval_ms": 337.52, "total_s": 0.7843867350020446, "ttft_s": 0.75572425199789 }, { "eval_count": 2, "prompt_eval_count": 354, "server_load_ms": 185.363719, "server_prompt_eval_ms": 358.775, "total_s": 0.8156643999973312, "ttft_s": 0.7872921620000852 }, { "eval_count": 2, "prompt_eval_count": 347, "server_load_ms": 194.923712, "server_prompt_eval_ms": 369.184, "total_s": 0.8549021529906895, "ttft_s": 0.8255536840006243 }, { "eval_count": 2, "prompt_eval_count": 349, "server_load_ms": 152.246329, "server_prompt_eval_ms": 356.92, "total_s": 0.7839786929980619, "ttft_s": 0.7554497230012203 } ] } }, "rapl_probe": [ "1790198338.077458050 package-0 79617154455 262143328850", "1790198338.084192049 core 194653307533 262143328850", "1790198338.090741296 uncore 7652943054 262143328850", "1790198338.097636480 psys 22641589620 262143328850" ], "rapl_unreadable_polls": 0, "seat_before": { "api_ps": { "models": [] }, "stamp": "2026-09-23T21:18:58Z" }, "serve_cgroup_memory_before": { "events": { "high": "0", "low": "0", "max": "43078", "oom": "0", "oom_group_kill": "0", "oom_kill": "0", "sock_throttled": "401" }, "pressure": [ "some avg10=1.67 avg60=0.40 avg300=0.08 total=15826057", "full avg10=1.67 avg60=0.40 avg300=0.08 total=15826057" ], "stat": { "anon": 29655040, "file": 1335656448, "file_mapped": 1769472, "shmem": 0 }, "swap_current": 2637824 }, "serve_unit_before": { "ActiveState": "active", "ControlGroup": "/user.slice/user-1000.slice/user@1000.service/app.slice/coi-ollama-laptop.service", "Environment": "TZ=UTC OLLAMA_HOST=127.0.0.1:withheld-2026-10-14 OLLAMA_KEEP_ALIVE=-1 OLLAMA_MAX_LOADED_MODELS=1 OLLAMA_=1 OLLAMA_FLASH_ATTENTION=1 OLLAMA_KV_CACHE_TYPE=f16 OLLAMA_NOPRUNE=1 OLLAMA_MODELS=/workshop/bench-coi-2026-09-23/store CUDA_VISIBLE_DEVICES= OLLAMA_VULKAN=0 GGML_VK_VISIBLE_DEVICES=-1 HIP_VISIBLE_DEVICES= ROCR_VISIBLE_DEVICES=", "ExecStart": "{ path=/workshop/bench-coi-2026-09-23/ollama-0.32.13/bin/ollama ; argv[]=/workshop/bench-coi-2026-09-23/ollama-0.32.13/bin/ollama serve ; ignore_errors=no ; start_time=[n/a] ; stop_time=[n/a] ; pid=0 ; code=(null) ; status=0/0 }", "MainPID": "1739347", "MemoryCurrent": "1404567552", "MemoryMax": "34359738368", "MemoryPeak": "30671613952", "NRestarts": "0" }, "show": { "capabilities": [ "completion", "vision", "tools", "thinking" ], "details": { "families": [ "qwen35" ], "family": "qwen35", "format": "gguf", "parameter_size": "27.8B", "parent_model": "", "quantization_level": "Q4_K_M" }, "model_info": { "general.architecture": "qwen35", "general.file_type": 15, "general.parameter_count": 27781427952, "general.quantization_version": 2, "qwen35.attention.head_count": 24, "qwen35.attention.head_count_kv": null, "qwen35.attention.key_length": 256, "qwen35.attention.layer_norm_rms_epsilon": 1e-06, "qwen35.attention.value_length": 256, "qwen35.block_count": 64, "qwen35.context_length": 262144, "qwen35.embedding_length": 5120, "qwen35.feed_forward_length": 17408, "qwen35.full_attention_interval": 4, "qwen35.image_token_id": 248056, "qwen35.mrope_sections": null, "qwen35.rope.dimension_count": 64, "qwen35.rope.dimension_sections": null, "qwen35.rope.freq_base": 10000000, "qwen35.rope.mrope_interleaved": true, "qwen35.rope.mrope_section": null, "qwen35.ssm.conv_kernel": 4, "qwen35.ssm.group_count": 16, "qwen35.ssm.inner_size": 6144, "qwen35.ssm.state_size": 128, "qwen35.ssm.time_step_rank": 48, "qwen35.ssm.v_head_reordered": true, "qwen35.vision.attention.head_count": 16, "qwen35.vision.block_count": 27, "qwen35.vision.embedding_length": 1152, "qwen35.vision.image_mean": null, "qwen35.vision.image_std": null, "qwen35.vision.longest_edge": 16777216, "qwen35.vision.num_channels": 3, "qwen35.vision.patch_size": 16, "qwen35.vision.shortest_edge": 65536, "qwen35.vision.spatial_merge_size": 2, "qwen35.vision.temporal_patch_size": 2, "qwen35.vision_end_token_id": 248054, "qwen35.vision_start_token_id": 248053, "tokenizer.ggml.add_eos_token": false, "tokenizer.ggml.add_padding_token": false, "tokenizer.ggml.eos_token_id": 248046, "tokenizer.ggml.eos_token_ids": null, "tokenizer.ggml.merges": null, "tokenizer.ggml.model": "gpt2", "tokenizer.ggml.padding_token_id": 248044, "tokenizer.ggml.pre": "qwen35", "tokenizer.ggml.scores": null, "tokenizer.ggml.token_type": null, "tokenizer.ggml.tokens": null }, "parameters": "presence_penalty 1.5\ntemperature 1\ntop_k 20\ntop_p 0.95", "system_sha256": null, "template_head": "{{ .Prompt }}", "template_sha256": "b507b9c2f6ca642bffcd06665ea7c91f235fd32daeefdf875a0f938db05fb315" }, "started": "2026-09-23T21:18:58Z", "tag": { "details": { "context_length": 262144, "embedding_length": 5120, "families": [ "qwen35" ], "family": "qwen35", "format": "gguf", "parameter_size": "27.8B", "parent_model": "", "quantization_level": "Q4_K_M" }, "digest": "7653528ba5cba4dd8e19da24aaddc7f4d0b5ecd93571c0825dfd4137958ec06e", "modified_at": "2026-09-23T21:18:57.602543232Z", "name": "qwen3.5:27b-q4_K_M", "size": 17420432728 }, "tasks": [ "c", "d" ], "tasks_run": { "c": { "board_idle_window_epoch": [ 1790198409.327, 1790198418.332 ], "board_task_window_epoch": [ 1790198419.36, 1790198482.375 ], "energy": { "cards": [ "" ], "decisions": 108, "joules": 5571.383783558607, "joules_per_decision": 51.58688688480192, "mean_watts": 88.41361222381417, "net_joules_per_decision": 38.340093461598684, "peak_watts": 111.33, "per_card": { "": { "joules": 5571.383783558607, "mean_watts": 88.41361222381417, "min_watts": 22.71, "peak_watts": 111.33, "samples": 64, "seconds": 63.015000104904175 } }, "samples": 64, "seconds": 63.015000104904175 }, "idle": { "cards": [ "" ], "joules": 204.44396759629248, "mean_watts": 22.70338312027723, "peak_watts": 22.9, "per_card": { "": { "joules": 204.44396759629248, "mean_watts": 22.70338312027723, "min_watts": 22.61, "peak_watts": 22.9, "samples": 10, "seconds": 9.005000114440918 } }, "samples": 10, "seconds": 9.005000114440918, "what": "the draw of these cards immediately before the arm, with the model resident and no work in flight" }, "posture_end": { "ac_online": "1", "battery": { "capacity": "93", "status": "Not charging" }, "boards_on_bus": [ "0, , NVIDIA GeForce RTX 5090 Laptop GPU, withheld-2026-10-14" ], "card": { "clocks.max.sm": "3090", "clocks.mem": "14001", "clocks.sm": "1590", "clocks_throttle_reasons.active": "0x0000000000000000", "driver_version": "595.91.07", "enforced.power.limit": "150.00", "memory.total": "24463", "memory.used": "16927", "name": "NVIDIA GeForce RTX 5090 Laptop GPU", "pci.bus_id": "withheld-2026-10-14", "pcie.link.gen.current": "5", "pcie.link.gen.max": "5", "pcie.link.width.current": "8", "pcie.link.width.max": "16", "persistence_mode": "Enabled", "power.default_limit": "95.00", "power.draw": "33.38", "power.limit": "[N/A]", "power.max_limit": "175.00", "pstate": "P0", "temperature.gpu": "45", "utilization.gpu": "0", "uuid": "", "vbios_version": "98.03.A2.00.21" }, "compute_apps": [ ", 55051, /workshop/bench-coi-2026-09-23/ollama-0.32.13/lib/ollama/llama-server, 16904 MiB" ], "fan": "unreadable on this board (fan.speed N/A); the operator's turbo fan per OPERATOR-CONDITIONS.md", "how": "nvidia-smi -i ; enforced.power.limit and clocks.max.sm READ, never typed; power.limit reads N/A on this board and the board refuses -pl", "loadavg": "2.85 3.32 2.89 1/1902 92324", "nvidia_powerd": "active", "stamp": "2026-09-23T21:21:24Z" }, "posture_start": { "ac_online": "1", "battery": { "capacity": "93", "status": "Not charging" }, "boards_on_bus": [ "0, , NVIDIA GeForce RTX 5090 Laptop GPU, withheld-2026-10-14" ], "card": { "clocks.max.sm": "3090", "clocks.mem": "14001", "clocks.sm": "1590", "clocks_throttle_reasons.active": "0x0000000000000000", "driver_version": "595.91.07", "enforced.power.limit": "150.00", "memory.total": "24463", "memory.used": "16915", "name": "NVIDIA GeForce RTX 5090 Laptop GPU", "pci.bus_id": "withheld-2026-10-14", "pcie.link.gen.current": "5", "pcie.link.gen.max": "5", "pcie.link.width.current": "8", "pcie.link.width.max": "16", "persistence_mode": "Enabled", "power.default_limit": "95.00", "power.draw": "22.84", "power.limit": "[N/A]", "power.max_limit": "175.00", "pstate": "P0", "temperature.gpu": "39", "utilization.gpu": "0", "uuid": "", "vbios_version": "98.03.A2.00.21" }, "compute_apps": [ ", 55051, /workshop/bench-coi-2026-09-23/ollama-0.32.13/lib/ollama/llama-server, 16892 MiB" ], "fan": "unreadable on this board (fan.speed N/A); the operator's turbo fan per OPERATOR-CONDITIONS.md", "how": "nvidia-smi -i ; enforced.power.limit and clocks.max.sm READ, never typed; power.limit reads N/A on this board and the board refuses -pl", "loadavg": "5.31 3.74 2.98 2/1923 80264", "nvidia_powerd": "active", "stamp": "2026-09-23T21:20:09Z" }, "rapl_idle_resident": { "domains": { "core": { "joules": 50.383904, "mean_watts": 6.977479815545145, "samples": 8, "seconds": 7.220931529998779, "wraps": 0 }, "package-0": { "joules": 105.714268, "mean_watts": 14.640395613917295, "samples": 8, "seconds": 7.220724821090698, "wraps": 0 }, "psys": { "joules": 1.958064, "mean_watts": 0.2711772358114138, "samples": 8, "seconds": 7.220606088638306, "wraps": 0 }, "uncore": { "joules": 0.187133, "mean_watts": 0.025917513465724882, "samples": 8, "seconds": 7.220329999923706, "wraps": 0 } }, "instrument": "RAPL via /run/rapl/energy (root sampler, 1 Hz; R-10d): psys = the platform's own reading (not wall); package-0 = the CPU package" }, "rapl_task": { "domains": { "core": { "joules": 1290.121636, "mean_watts": 20.845142179814882, "samples": 61, "seconds": 61.8907573223114, "wraps": 0 }, "package-0": { "joules": 1809.669928, "mean_watts": 29.239505208496382, "samples": 61, "seconds": 61.89126372337341, "wraps": 0 }, "psys": { "joules": 16.410542, "mean_watts": 0.26515270008754943, "samples": 61, "seconds": 61.890910387039185, "wraps": 0 }, "uncore": { "joules": 1.73907, "mean_watts": 0.02809901865296388, "samples": 61, "seconds": 61.89077353477478, "wraps": 0 } }, "instrument": "RAPL via /run/rapl/energy (root sampler, 1 Hz; R-10d): psys = the platform's own reading (not wall); package-0 = the CPU package" }, "seat_requests_task": { "gin_lines": 0, "how": "journalctl -u ollama (the system daemon :withheld-2026-10-14), read-only; a GIN line is one request it served", "lines": [], "since": "2026-09-23 21:20:19 UTC", "until": "2026-09-23 21:21:23 UTC" }, "summary": { "accuracy": 0.8888888888888888, "brier": 0.2222222222222222, "correct": 96, "errors": 0, "eval_tokens_total": 229, "floored_rows": 0, "letters_floored_total": 0, "median_seconds": 0.589994434005348, "n": 108, "n_labelled": 108, "n_reached": 108, "p95_seconds": 0.6460158600093564, "prompt_tokens_total": 31303, "refused": 0, "refused_by_reason": {}, "reliability": [ { "accuracy": null, "correct": 0, "gap": null, "hi": 0.1, "lo": 0.0, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.2, "lo": 0.1, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.3, "lo": 0.2, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.4, "lo": 0.3, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.5, "lo": 0.4, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.6, "lo": 0.5, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.7, "lo": 0.6, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.8, "lo": 0.7, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.9, "lo": 0.8, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": 0.8888888888888888, "correct": 96, "gap": 0.11111111111111116, "hi": 1.0, "lo": 0.9, "mean_p": 1.0, "n": 108, "sum_p": 108.0 } ], "unparsed": 0 }, "wall_seconds": 63.42712505599775 }, "d": { "board_idle_window_epoch": [ 1790198484.913, 1790198493.918 ], "board_task_window_epoch": [ 1790198494.946, 1790198536.958 ], "energy": { "cards": [ "" ], "decisions": 48, "joules": 3619.730680607557, "joules_per_decision": 75.41105584599077, "mean_watts": 86.15944715602166, "net_joules_per_decision": 55.05952665996472, "peak_watts": 108.14, "per_card": { "": { "joules": 3619.730680607557, "mean_watts": 86.15944715602166, "min_watts": 23.13, "peak_watts": 108.14, "samples": 43, "seconds": 42.01199984550476 } }, "samples": 43, "seconds": 42.01199984550476 }, "idle": { "cards": [ "" ], "joules": 209.38648211479187, "mean_watts": 23.252247084680853, "peak_watts": 23.35, "per_card": { "": { "joules": 209.38648211479187, "mean_watts": 23.252247084680853, "min_watts": 23.13, "peak_watts": 23.35, "samples": 10, "seconds": 9.004999876022339 } }, "samples": 10, "seconds": 9.004999876022339, "what": "the draw of these cards immediately before the arm, with the model resident and no work in flight" }, "posture_end": { "ac_online": "1", "battery": { "capacity": "93", "status": "Not charging" }, "boards_on_bus": [ "0, , NVIDIA GeForce RTX 5090 Laptop GPU, withheld-2026-10-14" ], "card": { "clocks.max.sm": "3090", "clocks.mem": "14001", "clocks.sm": "1590", "clocks_throttle_reasons.active": "0x0000000000000000", "driver_version": "595.91.07", "enforced.power.limit": "150.00", "memory.total": "24463", "memory.used": "16927", "name": "NVIDIA GeForce RTX 5090 Laptop GPU", "pci.bus_id": "withheld-2026-10-14", "pcie.link.gen.current": "5", "pcie.link.gen.max": "5", "pcie.link.width.current": "8", "pcie.link.width.max": "16", "persistence_mode": "Enabled", "power.default_limit": "95.00", "power.draw": "37.24", "power.limit": "[N/A]", "power.max_limit": "175.00", "pstate": "P0", "temperature.gpu": "46", "utilization.gpu": "0", "uuid": "", "vbios_version": "98.03.A2.00.21" }, "compute_apps": [ ", 55051, /workshop/bench-coi-2026-09-23/ollama-0.32.13/lib/ollama/llama-server, 16904 MiB" ], "fan": "unreadable on this board (fan.speed N/A); the operator's turbo fan per OPERATOR-CONDITIONS.md", "how": "nvidia-smi -i ; enforced.power.limit and clocks.max.sm READ, never typed; power.limit reads N/A on this board and the board refuses -pl", "loadavg": "2.03 3.01 2.81 3/1909 96945", "nvidia_powerd": "active", "stamp": "2026-09-23T21:22:18Z" }, "posture_start": { "ac_online": "1", "battery": { "capacity": "93", "status": "Not charging" }, "boards_on_bus": [ "0, , NVIDIA GeForce RTX 5090 Laptop GPU, withheld-2026-10-14" ], "card": { "clocks.max.sm": "3090", "clocks.mem": "14001", "clocks.sm": "1590", "clocks_throttle_reasons.active": "0x0000000000000000", "driver_version": "595.91.07", "enforced.power.limit": "150.00", "memory.total": "24463", "memory.used": "16927", "name": "NVIDIA GeForce RTX 5090 Laptop GPU", "pci.bus_id": "withheld-2026-10-14", "pcie.link.gen.current": "5", "pcie.link.gen.max": "5", "pcie.link.width.current": "8", "pcie.link.width.max": "16", "persistence_mode": "Enabled", "power.default_limit": "95.00", "power.draw": "23.27", "power.limit": "[N/A]", "power.max_limit": "175.00", "pstate": "P0", "temperature.gpu": "44", "utilization.gpu": "0", "uuid": "", "vbios_version": "98.03.A2.00.21" }, "compute_apps": [ ", 55051, /workshop/bench-coi-2026-09-23/ollama-0.32.13/lib/ollama/llama-server, 16904 MiB" ], "fan": "unreadable on this board (fan.speed N/A); the operator's turbo fan per OPERATOR-CONDITIONS.md", "how": "nvidia-smi -i ; enforced.power.limit and clocks.max.sm READ, never typed; power.limit reads N/A on this board and the board refuses -pl", "loadavg": "2.85 3.32 2.89 2/1902 92359", "nvidia_powerd": "active", "stamp": "2026-09-23T21:21:24Z" }, "rapl_idle_resident": { "domains": { "core": { "joules": 44.138803, "mean_watts": 5.346646919255279, "samples": 9, "seconds": 8.255417585372925, "wraps": 0 }, "package-0": { "joules": 109.024562, "mean_watts": 13.206123239188686, "samples": 9, "seconds": 8.255606889724731, "wraps": 0 }, "psys": { "joules": 2.17193, "mean_watts": 0.263103580707771, "samples": 9, "seconds": 8.255037784576416, "wraps": 0 }, "uncore": { "joules": 0.236023, "mean_watts": 0.028590119453321287, "samples": 9, "seconds": 8.255404472351074, "wraps": 0 } }, "instrument": "RAPL via /run/rapl/energy (root sampler, 1 Hz; R-10d): psys = the platform's own reading (not wall); package-0 = the CPU package" }, "rapl_task": { "domains": { "core": { "joules": 799.611357, "mean_watts": 19.8656568035067, "samples": 40, "seconds": 40.25093984603882, "wraps": 0 }, "package-0": { "joules": 1136.370076, "mean_watts": 28.231968626807166, "samples": 40, "seconds": 40.25118088722229, "wraps": 0 }, "psys": { "joules": 10.809299, "mean_watts": 0.268545654806443, "samples": 40, "seconds": 40.2512526512146, "wraps": 0 }, "uncore": { "joules": 1.061887, "mean_watts": 0.026381424047788777, "samples": 40, "seconds": 40.251314640045166, "wraps": 0 } }, "instrument": "RAPL via /run/rapl/energy (root sampler, 1 Hz; R-10d): psys = the platform's own reading (not wall); package-0 = the CPU package" }, "seat_requests_task": { "gin_lines": 0, "how": "journalctl -u ollama (the system daemon :withheld-2026-10-14), read-only; a GIN line is one request it served", "lines": [], "since": "2026-09-23 21:21:34 UTC", "until": "2026-09-23 21:22:17 UTC" }, "summary": { "accuracy": 0.9583333333333334, "brier": 0.08333333333333333, "correct": 46, "errors": 0, "eval_tokens_total": 96, "floored_rows": 0, "letters_floored_total": 0, "median_seconds": 0.892110426495492, "n": 48, "n_labelled": 48, "n_reached": 48, "p95_seconds": 0.9200688660057494, "prompt_tokens_total": 24550, "refused": 0, "refused_by_reason": {}, "reliability": [ { "accuracy": null, "correct": 0, "gap": null, "hi": 0.1, "lo": 0.0, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.2, "lo": 0.1, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.3, "lo": 0.2, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.4, "lo": 0.3, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.5, "lo": 0.4, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.6, "lo": 0.5, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.7, "lo": 0.6, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.8, "lo": 0.7, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.9, "lo": 0.8, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": 0.9583333333333334, "correct": 46, "gap": 0.04166666666666663, "hi": 1.0, "lo": 0.9, "mean_p": 1.0, "n": 48, "sum_p": 48.0 } ], "unparsed": 0 }, "wall_seconds": 42.17671504498867 } }, "temp_guard": { "max_temp_c": 56.0, "rule": ">= 87 C on 6 consecutive 5 s reads", "tripped": null }, "think": "off", "think_budget": 8, "util_before_pct": [ 0.0, 0.0, 0.0 ], "warm": { "done_reason": "length", "load_ms": 47114.40648, "runner_pids": [ 55051, 1739347 ], "seconds": 47.26087077999546 } }