{
 "arena_line": "cudaHostRegister PORTABLE ok; large pages skipped (STRATA_NO_LARGEPAGES); using 4 KB pages",
 "arena_pages": "4 KB pages (STRATA_NO_LARGEPAGES)",
 "borrow_slots": 2161,
 "end_status": {
  "phase": "done",
  "runs": 9,
  "t": 1791358857.292,
  "utc": "2026-10-07T07:40:57Z"
 },
 "engine_env": {
  "STRATA_NO_LARGEPAGES": "1"
 },
 "engine_env_readback": "the engine's arena line agrees with the env the config handed it",
 "engine_identity": {
  "cmake_cache": "/workshop/bench-flashnext/build/strata-6f32ec07/CMakeCache.txt",
  "cmake_home": "/workshop/bench-flashnext/strata-6f32ec07",
  "exe": "/workshop/bench-flashnext/build/strata-6f32ec07/strata",
  "exe_realpath": "/workshop/bench-flashnext/build/strata-6f32ec07/strata",
  "expected_commit": "6f32ec070f23ced9f50e704d854d775da52591ab",
  "git_head": "6f32ec070f23ced9f50e704d854d775da52591ab",
  "ok": true,
  "src": "/workshop/bench-flashnext/strata-6f32ec07",
  "why": "the build tree names the source tree, whose HEAD is the arm's commit"
 },
 "fences_held": [
  "netns",
  "unit_ip_filter"
 ],
 "identity": "9/9",
 "inner_pid": 3236853,
 "inner_rc": 0,
 "kind": "strata",
 "label": "R-S139-IQ3",
 "last_phase": "done",
 "placement": null,
 "residency": "whole",
 "row_id": "w2-r-s139-iq3",
 "runs": [
  {
   "ctx": "4K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "b42c6a181c3e02478e7e033585c5962f35eb046b680c0ae837ad8706a3c0674b",
   "prompt_tokens": 4038,
   "repeat": 0,
   "run": 1,
   "status": "ok",
   "tokens_per_round_derived": 2.753,
   "void_above_tpr_x_ceiling": 955.3,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "4K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "9744fee0d5b25fee2aef36ad0bc7c5c4611f73d11f4e6bd9e5b84b17d93d319b",
   "prompt_tokens": 4038,
   "repeat": 1,
   "run": 2,
   "status": "ok",
   "tokens_per_round_derived": 2.612,
   "void_above_tpr_x_ceiling": 906.4,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "4K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "214ec62ff2b1550ea59fc2fa7cc264e79ceed8841f5fd91e1709a8b73be2fa8f",
   "prompt_tokens": 4038,
   "repeat": 2,
   "run": 3,
   "status": "ok",
   "tokens_per_round_derived": 2.813,
   "void_above_tpr_x_ceiling": 976.1,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "32K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "507b7bc9961d968eaf07467339f314a3c76a036b2542852f07b9a638adcd5ad5",
   "prompt_tokens": 32038,
   "repeat": 0,
   "run": 4,
   "status": "ok",
   "tokens_per_round_derived": 2.349,
   "void_above_tpr_x_ceiling": 815.1,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "32K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "c8a1ccb2149bdf9595fe8d97b7183fd9efd61c32259e907606ddbac7710955b2",
   "prompt_tokens": 32038,
   "repeat": 1,
   "run": 5,
   "status": "ok",
   "tokens_per_round_derived": 2.438,
   "void_above_tpr_x_ceiling": 846.0,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "32K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "d6c964051c51baa9ef5cb009be0a921422c2b28350d2dbea610a6038552bdc18",
   "prompt_tokens": 32038,
   "repeat": 2,
   "run": 6,
   "status": "ok",
   "tokens_per_round_derived": 2.667,
   "void_above_tpr_x_ceiling": 925.4,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "128K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "2491e963527b8a39f7fff1aa2d04dab5eb4dd3ca12c7417c451f6216c665972c",
   "prompt_tokens": 128037,
   "repeat": 0,
   "run": 7,
   "status": "ok",
   "tokens_per_round_derived": 2.169,
   "void_above_tpr_x_ceiling": 752.6,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "128K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "94613594c1c300728a39aa8436592de0bdf88c05edeceb88cd9c6794de1fa097",
   "prompt_tokens": 128037,
   "repeat": 1,
   "run": 8,
   "status": "ok",
   "tokens_per_round_derived": 2.393,
   "void_above_tpr_x_ceiling": 830.4,
   "void_above_window_x_ceiling": 2084
  },
  {
   "ctx": "128K",
   "decode": "withheld (COUNT-UNVERIFIED): the laptop's llama-tokenize recount first; raw in R-S139-IQ3.run/inner-rows.jsonl",
   "identity": "ok",
   "prompt_sha256": "38d3c9a3177694d5b9006f1c813fddc9e6c5745048a59c545c41d199f780b364",
   "prompt_tokens": 128037,
   "repeat": 2,
   "run": 9,
   "status": "ok",
   "tokens_per_round_derived": 2.415,
   "void_above_tpr_x_ceiling": 838.0,
   "void_above_window_x_ceiling": 2084
  }
 ],
 "started": "2026-10-07T07:38:15Z",
 "state": "ended",
 "strata_line": {
  "pinned": [
   "--pack",
   "/workshop/bench-flashnext/packs-6f32ec07/iq3_s",
   "--native",
   "/workshop/bench-flashnext/models/IQ3_S/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf",
   "--ple-gguf",
   "/workshop/bench-flashnext/models/IQ3_S/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf",
   "--expert-profile",
   "/workshop/bench-flashnext/strata-6f32ec07/data/expert-profile.bin",
   "--expert-cache",
   "auto",
   "--prefill",
   "auto",
   "--spec",
   "4",
   "--spec-min-p",
   "0.5",
   "--mtp",
   "/workshop/bench-flashnext/mtp/rt-6f32ec07",
   "--max-context",
   "131072",
   "--kv",
   "int8"
  ],
  "pool_workers": "none passed (C-11); setup.py is never run, so its pool-worker rule is not read here; the engine's own worker line is the threads figure",
  "setup_rule_adds": [
   "--kv-resident",
   "32768"
  ],
  "setup_rule_on_this_box": [
   "--pack",
   "/workshop/bench-flashnext/packs-6f32ec07/iq3_s",
   "--native",
   "/workshop/bench-flashnext/models/IQ3_S/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf",
   "--ple-gguf",
   "/workshop/bench-flashnext/models/IQ3_S/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf",
   "--expert-profile",
   "/workshop/bench-flashnext/strata-6f32ec07/data/expert-profile.bin",
   "--expert-cache",
   "auto",
   "--prefill",
   "auto",
   "--spec",
   "4",
   "--spec-min-p",
   "0.5",
   "--mtp",
   "/workshop/bench-flashnext/mtp/rt-6f32ec07",
   "--max-context",
   "131072",
   "--kv",
   "int8",
   "--kv-resident",
   "32768"
  ],
  "this_box_mem_total_gib": 123.39
 },
 "thp": "enabled=madvise defrag=madvise",
 "written": "2026-10-07T07:41:05Z",
 "zero_doorbell_line": "strata verify: window up to 6 tokens, 74.7 MiB of device buffers (100% VRAM resident: zero-doorbell graph)"
}