AGENTS.md: document new env flags (RT_CASC/IIR4/KMAP/DELTA), chain wired status, capture proof

- Add RT_CASC, RT_IIR4_GEN, RT_IIR4_C/TAU/P/MULT/SR, RT_FIR, RT_FIR_Q, RT_KMAP_FC, RT_DELTA_DIST, RT_DELTA_STATE, RT_DBG_CASC to env flags table - Update chain_9_19 section: WIRED+GATED (was CRITICAL) - Add capture proof section: rendersnap2.py + /tmp/opencode/ datasets - Remove outdated PTRACE live-dump hazards (superseded by /proc/pid/mem capture)
Δ-rule: distance-aware gain wired (default flat), ph*.npz capture proof
2026-09-03 14:11:19 +03:00 · 2026-09-03 00:21:33 +03:00 · 2026-09-03 00:04:23 +03:00 · 2026-09-02 23:51:04 +03:00 · 2026-09-02 23:50:19 +03:00 · 2026-09-02 23:20:29 +03:00
359 changed files with 866152 additions and 1528 deletions
@@ -1,4 +1,8 @@
 # venv и тяжёлые/бинарные артефакты — в репо НЕ пушим
 # Подход: `*` игнорирует ВСЁ; правила !... возвращают нужные файлы.
 # ВАЖНО для git: шаблон `*` игнорирует и файлы, и директории. Чтобы git "зашёл"
 # в директорию, каждую директорию на пути нужно разыгнорировать (для nls_dasm —
 # handoff/ и handoff/nls_dasm/). Иначе `git add -n` откажет с "dir is ignored".
 *
 !**/*.py
 !**/*.txt
@@ -7,21 +11,46 @@
 !**/*.npy
 !**/*.npz
 !**/*.java
 !handoff/
 !**/*.cpp
 !**/*.hpp
 !**/*.c
 !**/*.h
 !**/*.lua
 !**/roadmap.md
 !**/AGENTS.md
 !.gitignore
 !dsp/
-# тяжеловесы / чувствительные (исключены даже с include выше)
+# --- расшифрованные дизассемблы декомпа: НУЖНО трекать (ценные артефакты) ---
-*.bin
+!handoff/
 !handoff/archive/
 !handoff/archive/*.md
 !handoff/nls_dasm/
 !handoff/nls_dasm/*.dis
 !handoff/nls_dasm/*.bin
 !handoff/*.md
 !handoff/*.py
 !handoff/*.json
 !handoff/*.npy
 # --- защитный харнесс валидации ---
 !scripts/
 # --- тяжёлые / бинарные / временные / инструменты: НЕ в git ---
 *.log
 *.wav
 *.rpp
 __pycache__/
 dsp/build/
 ghidra-proj/
 regions/
 regions_dep/
 regions_render/
 regions_rt/
 regions_rt2/
 soothe_rt.bin.regions/
-*.wav
+dl/
-*.rpp
+lib/
 bin/
 include/
 **/*.regions/
@@ -0,0 +1,133 @@
 # AGENTS.md — runbook for AI agents working in this repo
 > **Статус → `README.md:13` (единственный источник TOTAL).** Этот файл — только runbook: сборка, метрика, env-флаги, tooling hazard.
 > Полный журнал — `handoff/NOTES_LEVEL_INDEX.md` → `handoff/NOTES_LEVEL.md` (живая голова) + `handoff/archive/`.
 > Карта метода — `handoff/BLOCKMAP_529fe0.md`. ЭТОТ ФАЙЛ ЧИТАЙ ПЕРВЫМ.
 ## Золотое правило (обязательно)
 1. **Цель — bit-exact реверс кода**, НЕ эмпирическая подгонка кривых. Каждый параметр
   должен иметь источник (decomp адрес / live-таблица), а не быть подогнанным числом.
   Если берёшь эмпирику — явно пометь и занеси в «осталось».
 2. Источник истины — **декомпиляция** (`decomp_funs.txt`, `decomp_funs2.txt`,
   `handoff/nls_dasm/*.dis`) и **live-снимки** (`/tmp/snap_rt.bin`). Обновлять заметки
   при любом новом декоде.
 3. Команды через `rtk` (токен-фильтр), включая цепочки `&&`.
 ## Сборка и тесты (канон)
 ```bash
 # собрать библиотеку + framed_test (bridge, 44.1k) + render48k (структурная цепь, 48k/4096)
 cmake -S dsp -B dsp/build
 cmake --build dsp/build --target framed_test render48k
 # bridge-рендер (N=2048, hop=512, SR 44100, sqrt-Hann OLA)
 # формат полосы: fc,q,sens,level_scale  (одна полоса в comma-форме)
 ./dsp/build/framed_test /home/m/soothe-bt/tone1kq.wav /tmp/o.wav 678.7611083984375,0.99999785,12,600
 # многополосный: через отдельные аргументы fc q sens scale  (см. framed_test.cpp)
 # структурный рендер (внутренняя сетка 48000/4096: resample -> FUN_180529fe0 chain -> resample)
 ./dsp/build/render48k /home/m/soothe-bt/tone1kq.wav /tmp/o48.wav 1000,0.99999785,12
 # корпусные прогонки (62 случая, честная 24-bit метрика)
 python3 scripts/corpus.py                                            # bridge + guard --compare
 python3 scripts/corpus_structural.py                                 # структурная цепь
 python3 scripts/corpus_structural.py --vs-bridge scripts/baseline_bridge.json
 # отдельные модули (bit-exact черные проверки)
 cmake --build dsp/build --target twin_check tables_check fftconv_check vlog_check leveltrack_check levelpath_check fn529fe0_check exp2_check vlaw_check
 ./dsp/build/twin_check        # float-parity twin-резонатора
 ```
 > **TOOLING HAZARD**: cmake пропускает пересборку при изменении исходника в ту же секунду —
 > параметрические свипы молча гоняют STALE бинарь. Протокол: `touch` исходника перед сборкой +
 > проверять свежесть mtime бинаря (пример: /tmp/sweep_fresh.py паттерн в NOTES_LEVEL:21a).
 ### Метрика (tone-cmp, Goertzel stead-state)
 Сравнивать **тримкнутый** выход (длина = длина входа) с референсом `*_ref.wav` / `t1kq_b1f_*`:
 ```python
 # ref: t1kq_b1f_678.7611083984375.wav (или t1kq_b1f_1000.wav и т.д. по fc полосы)
 # вход: tone1kq.wav (тон 1000 Гц)
 import wave, numpy as np
 def load(p):
    w=wave.open(p,'rb'); n=w.getnframes(); d=w.readframes(n)
    return np.frombuffer(d,dtype=np.int16).astype(np.float64).reshape(-1,w.getnchannels()).mean(1)/32768
 def ta(x,f,sr=44100):
    x=x[-int(0.75*sr):]; t=np.arange(len(x))/sr; w=2*np.pi*f
    return np.hypot(2*np.sum(x*np.cos(w*t))/len(x), 2*np.sum(x*np.sin(w*t))/len(x))
 def db(a): return 20*np.log10(max(a,1e-9))
 inp=load('/home/m/soothe-bt/tone1kq.wav')
 out=load('/tmp/o.wav')
 ref=load('/home/m/soothe-bt/t1kq_b1f_678.7611083984375.wav')   # fc полосы = 678.76
 err = db(ta(out,1000)/ta(ref,1000))     # err в dB, цель <0.1
 ```
 Тестовый корпус (`/home/m/soothe-bt/`, вне git): `tone1kq.wav`(вход),
 `t1kq_b1f_<fc>.wav` (рефы), `comb.wav`/`comb_ref.wav` (мультиполосный).
 ## Структура ключевых файлов
 - `dsp/framed_model.{cpp,hpp}` — C++ порт mask-apply цепи (ГЛАВНЫЙ файл).
 - `dsp/render48k.cpp` — пайплайн 48000/4096 (resample → chain → resample).
 - `dsp/twin.{cpp,hpp}` — twin FUN_180535880 (float-parity).
 - `scripts/corpus.py` / `corpus_structural.py` — харнессы + guard `--compare`.
 - `handoff/NOTES_LEVEL.md` — живой журнал (голова 24mm5+, хвост → `archive/`).
 - `handoff/NOTES_LEVEL_INDEX.md` — оглавление журнала.
 - `handoff/BLOCKMAP_529fe0.md` — блок-карта FUN_180529fe0.
 - `handoff/nls_dasm/` — 134 дизассембла.
 ## Env-флаги экспериментов (актуальные)
 | флаг | действие |
 |------|----------|
 | `RT_VLAW=1` | закон применённой стадии: `mask=10^(−(α·ln1p(lvl/β)+c)/20)` |
 | `RT_SYN=1` | STFT БЕЗ синтез-окна (найденный слой плагина) |
 | `RT_WIN=0/1/2` | окно анализа: sym-hann / periodic / rect |
 | `RT_NOWARP=1` | отключить warp-модуляцию маски |
 | `RT_NOIIR3=1` | отключить IIR3 ×2 |
 | `RT_IIR12=0` | отключить частотные IIR1/2 (КРИТИЧНО с RT_VLAW — иначе размывают дипы) |
 | `RT_DUMP_BIN=<f>` (+`RT_DUMP_FRAME=N`) | дамп тракта бина N: am/res/lvl_raw/band_level/prewarp/w |
 | `RT_VDBG=1` | stderr-печать vlaw-вычислений |
 | `RT_FAITHFUL=1` | faithful-цепь `dsp/fnfaith.cpp` (детекторный каскад) |
 | `RT_FIRCONV=1/3` | FIR-применение (1=complex-mul, 3=`1.019·mask^1.8345`) |
 | `RT_CASC=1` | цепь 9–19 (`chain_9_19`, IIR4×2 double + FIR min-phase, gated) |
 | `RT_IIR4_GEN=1` | генератор IIR4 (FUN_180533340) vs proxy `kIIR_A1/B1` |
 | `RT_IIR4_C/TAU/P/MULT/SR` | параметры генератора IIR4 (1000/1200/0.5/360/48000) |
 | `RT_FIR=0/1` | пуск FIR min-phase (default 1) |
 | `RT_FIR_Q=x` | показатель FIR (default 0.8002203702926636, live `.rdata`) |
 | `RT_KMAP_FC=1` | fc-фактор k-mapping через W_eq (default off) |
 | `RT_DELTA_DIST=x` | Δ distance factor (template-local gain, default 0=flat) |
 | `RT_DELTA_STATE=1` | STATE-dependent Δ (опен-ин, wired, empty) |
 | `RT_DBG_CASC=1` | gate DBG_CASC fprintf (per-frame spam otherwise) |
 Полный набор dual-решения: `RT_VLAW=1 RT_SYN=1 RT_NOWARP=1 RT_NOIIR3=1 RT_IIR12=0`.
 ## Инструменты
 | скрипт | назначение |
 |--------|-----------|
 | `scripts/rendersnap2.py <rpp> [cap] [outdir]` | мягкий STOP-снаппер (RENDER_FILE из rpp!) |
 | `scripts/campaign.py <base> <fc> <q> <sens> <in> <drives> <out>` | ячейка параметризации (~8 мин) |
 | `scripts/disasm_func.py <VA> [len]` | capstone-дизасм с RIP-константами |
 | `scripts/iat_name.py` | рантайм-резолв импортов (SIGSTOP!) |
 | `scripts/probe_states.py` / `probe_mem.py` / `dump_dispatch.py` | live state/память/таблицы |
 | `scripts/hunt2.py` | перебор ctx-инстансов |
 | `scripts/scan_pairs.py`, `scan_lutsub.py` | диагностика памяти |
 Датасеты: `/tmp/opencode/sc_{q,sens,qmap,k,f,d}*` + `tract_*` + `*.pkl/.npz` (см. `NOTES_LEVEL_INDEX.md`).
 Открытые пробелы → `BITEXACT_PLAN.md`.
 ## Архитектурная проблема chain_9_19 (WIRED, GATED)
 `chain_9_19` из `dsp/fn52fe0.cpp:259` (BLOCKMAP:620-644) — per-frame цепь с **persistent ACC state** (`@ctx+0x5407c8`). Интегрирована в `framed_model.cpp:process_band_structural` через `RT_CASC=1`. IIR4×2 использует double precision (movsd/mulsd per disasm 1191), FIR min-phase бит-точен до df0 (0.0065 dB). Генератор IIR4 (`generate_iir4_coefs`, FUN_180533340) env-gated `RT_IIR4_GEN`.
 **Статус:** chain wired и gated (`RT_CASC=0` default = canon untouched). Все unit checks PASS, corpus gate d=+0.000. Калибровка I/O format требует `ph*.npz` capture (rendersnap2.py) — capture proof получен (2026-09-02).
 **Live-dump chain (ptrace):** chain вызывается в рантайме (DIV#0-7 на `0x1803a06a0`):
 - Input: `a` = bands_curve (VLAW output, min=0, max=17.6, mean=0.048)
 - `b` = tmp6f8 (step 9b accumulation, min=0, max=0.8, mean=0.8)
 - 8 DIV hits на одном кадре (dual-band: 2 bands × 4 iterations?)
 **Capture proof (2026-09-02):** `rendersnap2.py` снимает `ph*.npz` (слоты 0x540628/0x540678/0x540688/0x540768) через `/proc/pid/mem` без ptrace-брейкпоинтов. Данные в `/tmp/opencode/{rendersnap2_dual,snap_t1k_b1f_1000,snap_dual300,rendersnap2_comb_b1234}`. Маск-цепь работает в рендер-окне (BLOCKMAP:285 устарел).
 ---
 ## Чистая работа
 - Не коммитить: `*.bin`, `*.wav/rpp`, `*.log`, `dsp/build/`, `ghidra-proj/`, `dl/lib/bin/include/`, `regions*/`. См. `.gitignore`.
 - После правки C++ — `touch` исходника + `cmake --build dsp/build --target framed_test` (hazard `AGENTS.md:66`) + `python3 scripts/corpus.py --compare scripts/baseline_bridge.json --tol 0.25`.
@@ -0,0 +1,90 @@
 # BIT-EXACT PLAN — путь от dB-приближения к побайтовой парности
 Статус: **НЕ bit-exact**. Канон — bridge `framed_test` 44.1k: **TOTAL 1.594 dB** (62 случая, trimmed 24-bit, `scripts/baseline_bridge.json`). Bit-exact достижим (F0 gate: рендеры байт-детерминированы `handoff/NOTES_LEVEL.md:1020`), но модель — погона. Читать вместе с `AGENTS.md` и `handoff/NOTES_LEVEL_INDEX.md`.
 > **2026-08-28 (24mm14): применение декодировано до формул.** `mask=10^(−cut_D/20)`, `cut_D=α·ln1p(lvl/β)+c [+Δ]` — три семейства, rms ≤0.016. `dual` 0.193 dB (env-gated `RT_VLAW=1 RT_SYN=1 RT_NOWARP=1 RT_NOIIR3=1 RT_IIR12=0`), детекторный каскад `180529c60` + RFFT до df0 бит-точны, `γ=1.760561` из цепи. **Гейт смены канона = bit-exact** (решение `24hh2`). Остаток: 3 шага ниже.
 Цель фазы C: воспроизвести `FramedDetector` (= `FUN_180529fe0` mono-path) до `verify_bit_exact.py` побайтового совпадения (`t1kq_*`, `dual_*`, `comb_*`).
 ---
 ## 0. Текущее состояние — ЕДИНСТВЕННЫЙ ИСТОЧНИК `README.md:13`
 Честная trimmed метрика 62 случая (bridge vs structural 48k/4096 `corpus_structural.py`):
 | Группа | bridge mean | bridge max | structural mean | VLAW env-gated mean |
 |--------|-------------|------------|-----------------|---------------------|
 | t1kq (fc-scan) | 0.226 | 0.500 | 0.768 | 0.426–0.852* |
 | t1k (loud) | 1.801 | 2.227 | 2.114 | 0.577–0.930* |
 | al (level) | 0.638 | 1.596 | 0.986 | 0.090–0.804* |
 | res | 0.628 | 1.535 | **0.437** | 0.395* |
 | dual (q-sweep) | 0.726 | 2.252 | 3.264 | **0.193** (max 0.438) ✓ |
 | comb (4-band) | **10.149** | 14.752 | **6.116** | 2.678–4.335* |
 | **TOTAL** | **1.594** | 14.752 | 2.286 | **0.870–1.931*** |
 `*` зависит от калибровки VLAW `α/β/c` (семейство dual vs fc-зависимая `AGENTS.md` больше не хранит таблицу — см. `README.md:13` и `handoff/NOTES_LEVEL.md:3919`).
 Канон = bridge `1.594` (эмпирика `C=G·LUT(xv)+W·warp^A, gain=(1−C)·res^rp`). Structural `2.286` обгоняет bridge на `comb/res`, регрессирует на однополосных — форма кривой редукции неверна. VLAW закрывает `dual` (0.193) — применённый слой и каскад декодированы.
 ---
 ## 1. Порядок работ (3 шага, обязательный порядок; каждый валидируется отдельно)
 > **Валидация:** `cmake --build dsp/build --target framed_test` + `python3 scripts/corpus.py` + `python3 scripts/corpus.py --compare scripts/baseline_bridge.json --tol 0.25` (фейлит при регрессе группы >tol). Структурная цепь валидируется `scripts/corpus_structural.py --vs-bridge`. Каждый под-шаг коммитить отдельно.
 ### Шаг 1 — Каскадный симулятор шагов 9–19 FUN_180529fe0 (ПРИОРИТЕТ №1)
 Dataflow декодирован `handoff/BLOCKMAP_529fe0.md:22v` + `handoff/NOTES_LEVEL.md:4400`:
 ```
 bands[i] → LOG#1 (140980) → 52d650 (bidir #1) → vec698*= (1−[54087c]) → vec6f8+= [54087c]·0.8
 → DIVIDE vec6f8/bands (1803a06a0) → vec6f8=bands−ACC (dc40) → fma ATT/REL (1fa0/1940, тройки re,im,coef)
 → COPY (1b80) → mirror 9 + bigkernel 140b60 → EXP#1 (1409e0) + −1 → ×track (th2000) → ×warp (th2000)
 → LOG#2 (140980) → IIR4×2 (bidir, log-домен) → EXP#2 → bands_final
 → FIR-секция: DESIGN 1802a24c0 (conv_float_a24c0.dis) → copy → RFFT pair th2180/th1a90 (buf548/598)
  → FIR[1..n/2]*=2.0 → EXP 1803831c0 → WINfreq· → FIR[0]=1 → df0 track⊗FIR → mask
 ```
 Тела bigkernel'ов резолвлены статически `handoff/NOTES_LEVEL.md:3944` (`1803a06a0` divide, `180296c80` expf, `1803831c0` табличная кривая, `1803a06a0` etc.; `24mm2` таблица). RFFT до df0 бит-точна `24mm11`.
 **Делать:** оп-за-оп транскрипция `scripts/cascade_sim.py` (numpy-эквиваленты divide/expf с FMA-точностью) → валидация на `multi6` (6 нотчей один прогон, `handoff/NOTES_LEVEL.md:3809`) и `sc_*` дистанционной серии (rms сейчас 0.42 на угаданных формах). После структуры — C++ порт `dsp/fn529fe0.cpp` с точными полиномами.
 Критерий: каскад воспроизводит `B=exp(scratch)` из `multi6` на всех 6 пиках ±0.05 dB (сейчас `rms 0.42`), затем `dual` 0.193 сохраняется без per-семейной калибровки.
 ### Шаг 2 — k-маппинг фронтенда (twin/am)
 `k = lvl_impl / lvl_ours` (`handoff/NOTES_LEVEL.md:3436`): `k(q≥2)=0.403` const, `k(sens)` экспонента (5.37 при s18, 22.0 при s24). При q=1.0 vs q=0.5 `k` идентичен — `q` НЕ влияет на закон (`24kk`), `sens` линейно через `lvl_raw` (`24kk`). Остаток — twin gain пол (`res` плагина ≥0.153 vs наш 0.0069) и амплитудная нормировка `am` (wsum/окно в `dsp/spectral.cpp` vs декомп `FUN_180535880` tail-calls).
 **Делать:** декод `FUN_180535880` tail-calls + `FUN_180536300` caller (grid `NOTES_TWIN.md`) до точных `twin_coeff`/`am` формул, либо калибровка `k(q,sens,fc)` по `sc_*` + `tract_*` как таблицы с декомп-обоснованием.
 Критерий: cross-config `α/β/c` становятся вычисляемыми, а не фитовыми; `sens 6..24` перестаёт требовать per-групповой `α`.
 ### Шаг 3 — Δ-правило вторых пиков (pre-combine)
 Второй контент-пик: `cut = α·ln1p(g·lvl/β)+c`, `g@2000≈12.15` vs `g@fc=1.0` (`handoff/NOTES_LEVEL.md:3890`), но `g` меняется между прогонами (12.15 vs 1.85 при той же геометрии) — зависит от `STATE=[ctx+0x540788]` (спектральная память, `24ii2`). Смешение шаблонно-локальное (`24ll`: далёкий тон@4000 не влияет), ширина провалов константа `24k-2`.
 **Делать:** серия `campaign.py` с вариацией `STATE` (разный контент до основного тона) при фиксированной геометрии → формула `g(b)=G(геометрия, STATE)` + pre-combine шага `52a397` (`handoff/BLOCKMAP_529fe0.md:43`).
 Критерий: `d1500` серия `3.43→4.23` монотонна без отрицательных `w(d)` (`24mm`).
 ---
 ## 2. Что НЕ делать
 - **НЕ вводить эмпирию где есть декомп** (золотое правило `AGENTS.md:33`). `LCP` и `MULT` без источника — помечать `EMPIRICAL`.
 - **НЕ делать combine в bridge-final-gain домене** — F2 регресс 4.68/5.22 `handoff/archive/NOTES_LEVEL_2026-08-18_2026-08-23.md:1064`.
 - **НЕ читать WAV наивно** — только `render_parity.load` (bext/junk, 24-bit `handoff/NOTES_LEVEL.md:1462`).
 - **НЕ править XML `<PARAM>` в RPP** — декоративная копия, стейт в бинарной части (`setparam.lua`/`patchparam.py`).
 - **НЕ свёртка/OLA поверх маски** — опровергнуто `24j` (`A=1.019·V^1.8345` = произведение экспонент, не свёртка).
 ## 3. Риски и время
 - Доминирующий риск — каскад 9–19: 5700 строк AVX-512+FMA (`1803831c0`), 10 ядер. Митигация: numpy-симулятор до C++ порта, валидация на `multi6` самосогласованно.
 - Twin/am — multi-week (dispatch ядра 2/4/8, plan-gen `FUN_18002f980`). Не блокирует Шаг 1.
 - Стерео M8 — отложен (все рендеры mono).
 - Оценка: Шаг 1 — 1–2 недели, Шаг 2 — 1–2 недели, Шаг 3 — дни. До байтов — 1–2 месяца.
 ## 4. Точка входа
 1. `AGENTS.md` → `handoff/NOTES_LEVEL_INDEX.md` (живая голова `24mm5+`) + `handoff/BLOCKMAP_529fe0.md`.
 2. `touch dsp/framed_model.cpp && cmake --build dsp/build --target render48k framed_test` (hazard `AGENTS.md:66`).
 3. Бейзлайны: `python3 scripts/corpus.py --compare scripts/baseline_bridge.json --tol 0.25` (TOTAL 1.594); VLAW dual `RT_VLAW=1 RT_SYN=1 RT_NOWARP=1 RT_NOIIR3=1 RT_IIR12=0` → 0.193.
 4. Инструменты: `scripts/cascade_sim.py` (структурная фаза), `scripts/rendersnap2.py`, `scripts/disasm_func.py`, `scripts/iat_name.py`, `scripts/campaign.py`.
@@ -0,0 +1,206 @@
 import ghidra.app.script.GhidraScript;
 import ghidra.app.decompiler.DecompInterface;
 import ghidra.app.decompiler.DecompileResults;
 import ghidra.program.model.listing.Function;
 import ghidra.program.model.listing.FunctionManager;
 import ghidra.program.model.listing.Instruction;
 import ghidra.program.model.address.Address;
 import ghidra.program.model.address.AddressSpace;
 import ghidra.program.model.symbol.Reference;
 import ghidra.program.model.mem.Memory;
 import java.io.PrintWriter;
 import java.io.FileWriter;
 import java.io.BufferedReader;
 import java.io.FileReader;
 import java.util.*;
 /**
 * DumpFuns2 - unfrozen BFS decompiler.
 * Differences vs DumpFuns.java:
 *   1. NO haveDec gate: always decompiles into decomp_funs2.txt.
 *   2. Extended seeds: explicit critical DSP addresses + reverse callers
 *      (references-to) so indirect/vtable-dispatched functions enter closure.
 *   3. Skip threshold lowered from 30 to 8 instructions (keeps dispatchers/stubs).
 * Also writes fun_map2.txt (fresh) and re-scans .data float constants.
 */
 public class DumpFuns2 extends GhidraScript {
    final long LO = 0x180000000L;
    final long HI = 0x182a00000L;
    long parseAddr(String s) {
        try {
            return Long.parseLong(s.replaceAll("[^0-9a-fA-F]", ""), 16);
        } catch (Exception e) { return -1; }
    }
    boolean inRange(long a) { return a >= LO && a <= HI; }
    @Override
    public void run() throws Exception {
        FunctionManager fm = currentProgram.getFunctionManager();
        AddressSpace as = currentProgram.getAddressFactory().getDefaultAddressSpace();
        // --- seeds: all FUN_ from decomp_dsp.txt + explicit critical DSP addrs ---
        Set<Long> seed = new LinkedHashSet<>();
        BufferedReader br = new BufferedReader(new FileReader("/home/m/re-tools/decomp_dsp.txt"));
        String line;
        while ((line = br.readLine()) != null) {
            int i = line.indexOf("FUN_");
            while (i >= 0) {
                int e = i + 4;
                while (e < line.length() && "0123456789abcdefABCDEF".indexOf(line.charAt(e)) >= 0) e++;
                long a = parseAddr(line.substring(i + 4, e));
                if (inRange(a)) seed.add(a);
                i = line.indexOf("FUN_", e);
            }
        }
        br.close();
        long[] critical = {
            0x180529fe0L, // mask/coefficient setup (vtable-only, 2051 instr)
            0x180563440L, // LUT curve + gamma + combine
            0x180563ce0L, // IIR level-tracker INIT
            0x18056e3e0L, // twin-mask factory
            0x18052f500L, // interleave
            0x18052b550L, // FFT-conv loop
            0x18052ee70L, // per-bin gain
            0x18052d650L, // band setup
            0x18052d920L, // window
            0x18052e190L, // buffer alloc
            0x18052dc30L, // FFT plan builder
            0x18052da00L, // ramp fill
            0x18052dbc0L, // cplx-interleave
            0x180535880L, // twin kernel
            0x180536300L, // twin caller
            0x180536f90L, // dead sibling
            0x180530d30L, // level-weight formula
            0x1805316e0L, // coeff writer (17 case)
            0x180563fa0L, // IIR level-tracker UPDATE (caller chain 5631c0)
            0x1805631c0L, // LUT+IIR-init+UPDATE dispatcher
            0x180563260L, // param dispatcher
            0x180563a60L  // combine (called from 563440)
        };
        for (long a : critical) seed.add(a);
        // --- vtable slots ---
        long[] vts = {
            0x1824abb90L, 0x1824ac7a8L, 0x1824ab7c0L, 0x1824ac638L, 0x1824ac5d8L,
            0x1824be810L, 0x1824be228L, 0x1824be0a8L, 0x1824be7a0L, 0x1824ac210L,
            0x1824ac248L, 0x1824b1ac8L, 0x1824b11c8L, 0x1824b0fd0L, 0x1824b1178L
        };
        for (long base : vts) {
            for (int i = 0; i < 64; i++) {
                try {
                    Address va = as.getAddress(base + i * 8L);
                    long tgt = currentProgram.getMemory().getLong(va);
                    if (inRange(tgt)) seed.add(tgt);
                } catch (Exception e) { }
            }
        }
        // --- BFS closure forward + REVERSE (indirect callers) ---
        Set<Long> done = new TreeSet<>();
        Deque<Long> queue = new ArrayDeque<>(seed);
        List<Long> order = new ArrayList<>();
        int cap = 8000;
        while (!queue.isEmpty() && order.size() < cap) {
            long a = queue.poll();
            if (!done.add(a)) continue;
            order.add(a);
            Function f = fm.getFunctionAt(as.getAddress(a));
            if (f == null) continue;
            try {
                for (Function c : f.getCalledFunctions(monitor)) {
                    long ca = c.getEntryPoint().getOffset();
                    if (inRange(ca) && !done.contains(ca)) queue.add(ca);
                }
            } catch (Exception e) { }
            // reverse: any function referencing this entry (indirect dispatch callers)
            try {
                for (Reference r : currentProgram.getReferenceManager()
                        .getReferencesTo(as.getAddress(a))) {
                    Address from = r.getFromAddress();
                    if (from == null) continue;
                    Function cf = fm.getFunctionContaining(from);
                    if (cf != null) {
                        long ca = cf.getEntryPoint().getOffset();
                        if (inRange(ca) && !done.contains(ca)) queue.add(ca);
                    }
                }
            } catch (Exception e) { }
        }
        // --- fun_map2.txt fresh ---
        PrintWriter pm = new PrintWriter(new java.io.BufferedWriter(
            new FileWriter("/home/m/re-tools/fun_map2.txt", false)));
        for (long a : order) {
            Function f = fm.getFunctionAt(as.getAddress(a));
            if (f == null) continue;
            pm.println(Long.toHexString(a) + " " + f.getBody().getNumAddresses() + " " + f.getName());
        }
        pm.close();
        // --- decompile ALWAYS into decomp_funs2.txt ---
        DecompInterface di = new DecompInterface();
        di.openProgram(currentProgram);
        PrintWriter pw = new PrintWriter(new java.io.BufferedWriter(
            new FileWriter("/home/m/re-tools/decomp_funs2.txt", false)));
        int n = 0, skipped = 0;
        for (long a : order) {
            Function f = fm.getFunctionAt(as.getAddress(a));
            if (f == null) continue;
            long sz = f.getBody().getNumAddresses();
            if (sz < 8) { skipped++; continue; }
            DecompileResults res = di.decompileFunction(f, 120, monitor);
            if (res != null && res.getDecompiledFunction() != null) {
                n++;
                pw.println("############ FUN_ " + Long.toHexString(a) + " size=" + sz + " ############");
                pw.println(res.getDecompiledFunction().getC());
                pw.println();
            }
        }
        pw.close();
        di.dispose();
        println("DECOMP2_DONE n=" + n + " skipped=" + skipped + " order=" + order.size());
        // --- constants (same as DumpFuns, scaled to what we touched) ---
        Map<Long,String> consts = new TreeMap<>();
        Memory mem = currentProgram.getMemory();
        for (Function f : fm.getFunctions(true)) {
            long fa = f.getEntryPoint().getOffset();
            if (!order.contains(fa)) continue;
            Iterator<Instruction> iit = currentProgram.getListing().getInstructions(f.getBody(), true);
            while (iit.hasNext()) {
                Instruction ins = iit.next();
                for (Reference r : currentProgram.getReferenceManager().getReferencesFrom(ins.getAddress())) {
                    long ta = r.getToAddress().getOffset();
                    if ((ta >= 0x1824c0000L && ta <= 0x182700000L) ||
                        (ta >= 0x180000000L && ta <= 0x181200000L)) {
                        consts.merge(ta, Long.toHexString(fa), (x, y) -> x + ";" + y);
                    }
                }
            }
        }
        PrintWriter pc = new PrintWriter(new java.io.BufferedWriter(
            new FileWriter("/home/m/re-tools/consts2.txt", false)));
        for (Map.Entry<Long,String> e : consts.entrySet()) {
            try {
                long ta = e.getKey();
                int[] b = new int[8];
                boolean ok = true;
                for (int k = 0; k < 8; k++) {
                    Address a = as.getAddress(ta + k);
                    if (!mem.getLoadedAndInitializedAddressSet().contains(a)) { ok = false; break; }
                    b[k] = mem.getByte(a) & 0xff;
                }
                if (!ok) continue;
                float f32 = Float.intBitsToFloat(b[0] | (b[1] << 8) | (b[2] << 16) | (b[3] << 24));
                float f32b = Float.intBitsToFloat(b[4] | (b[5] << 8) | (b[6] << 16) | (b[7] << 24));
                pc.println(Long.toHexString(ta) + " f32=" + f32 + " f32b=" + f32b + "  refs=" + e.getValue());
            } catch (Exception ex) { }
        }
        pc.close();
        println("CONSTS2_DONE consts=" + consts.size());
    }
 }
@@ -1,93 +1,111 @@
 # soothe2-re
 Обратный инжиниринг DSP-ядра **oeksound soothe2** (VST3, Windows x64) → проверяемая
-численная модель и реконструкция на C++.
+реконструкция на C++ с **bit-exact** целью.
-Цель — понять, как именно плагин считает подавление резонансов, и воспроизвести
+Цель — понять, как именно плагин считает подавление резонансов (уровневый детектор,
-его поведение (в идеале бит-точно, но пока — с точностью ~0.03–0.18 dB на формах
+маска, фильтр), и воспроизвести это дословно. Текущий активный канон — C++
-полос и уровнях).
+`FramedDetector` (`dsp/framed_model.cpp`): real mask-apply цепь с live-таблицами.
 Канонные команды и метрика — в [`AGENTS.md`](AGENTS.md).
 ---
-## Статус (B.15, август 2026)
+## Статус (2026-09-02, chain919) — ЕДИНСТВЕННЫЙ ИСТОЧНИК TOTAL
-**Полный STFT-рендер** пайплайна работает в `framed_render.py` (N=2048, hop=512,
+**Bridge (`framed_test` 44.1k): TOTAL 1.594 dB** (`scripts/baseline_bridge.json`, guard `--tol 0.25`). **Структурный `render48k` 48k/4096 L/R (`RT_VLAW=1 RT_SYN=1 RT_NOWARP=1 RT_NOIIR3=1 RT_IIR12=0`): TOTAL 2.689 dB** (dual 2.04, t1k 5.43, max `6.8`). Исторический канон `0.341` (24mm14, `f40f41e`) недостижим на текущем `HEAD` (L/R `e343b0a` + chain gated `RT_CASC=0`); требует перекалибровки `per-fc`/`EQ`. Guard: `python3 scripts/corpus_structural.py` + `python3 scripts/corpus.py --compare scripts/baseline_bridge.json --tol 0.25`. Chain 9–19 интегрирован (`RT_CASC=1` gated, IIR4 double + FIR min-phase) но некалиброван (`31.6` при включении, блокер — входной формат level vs cut). Следующий шаг — калибровка chain по live-dump/`ph*.npz` (BLOCKMAP:285). Детали — `handoff/NOTES_LEVEL_INDEX.md`, `.opencode/plans/chain919.md`.
 sqrt-Hann, per-bin twin-envelope tatt=11ms/trel=80ms). Модель:
 `C(f_k)=g·LUT(log10(A_k/res_k)) + w·warp(f_k)^a`, `gain=(1−C)·res^(rp0·Q^drp)`.
- **Честная (trimmed) метрика** — длина выхода = длина входа (не-тримнутые замеры
+**Цель — bit-exact** (гейт = все параметры до декомпа + корпус в шумовой пол). Декомп ~95%, `handoff/nls_dasm/` ~140 `.dis`.
-  давали ложный сдвиг ~0.2 dB, исправлено в B.14). Базлайн для scalar-rp = 0.280 dB.
+
- **Q-dependent rp** `rp(Q)=rp0·Q^drp` (rp0=0.0275, drp=0.2159) → **mean=0.175 dB**:
+**Применение декодировано**:
-  q0.1 и q10 почти идеальны, боттлнек q1@2000 (−0.71 dB).
+```
- **JOINT free-knot LUT** (8 узлов Pchip, G/W/A/rp0/drp) → **dual mean≈0.027 dB**,
+mask(b)   = 10^(−cut_D(b)/20)                    ← вещественная, per-bin multiply
-  q1@2000 закрыт до 0.000; остаток al_* lv24 dC=+0.10 (зона xv<−0.3).
+cut_D(b)  = α·ln(1+lvl_raw EQ(b)/β)+c [+Δ content-aware, lvl EQ before detector]
- **Декомп DSP-ядра в основном закрыт**: twin-резонатор, генератор case8, level-weight
+lvl_raw   = am/res·scale · W_eq(fc,q,sens) (W=10^(sens·H·0.3/12/20), H=1/√(1+(1.54q^1.33·A)²))
-  (0x530d30 — численно НЕ create tilt, подтверждено при N=2048), LUT-кривая FUN_180563440
+слой      = STFT БЕЗ синтез-окна (RT_SYN=1), EQ до детектора (pipeline_ocr)
-  (linear/power-law по флагу), IIR-трекеры FUN_180563ce0, FFT-conv 0x535a70 —
+```
-  всё размаплено, `.dis` в `handoff/nls_dasm/`.
+Калибровки: α/β/c контент-зависимы (dual `3.22/0.49/0.54 Δ6.92` rms 0.016), q не влияет (`24kk`), sens линейно, `fc`-геом через `W_eq` (`24mm14` `R bandpass`), `dual q1.0` misclass пофикшено `has_second_peak+maxlvl>2` (`dsp/framed_model.cpp:331`), `t1k_500` `4.5/0.35` + `comb 0.05/5.0`.
 - **dual+EQ**: исторический `0.341` (канон `f40f41e`) сейчас `2.689` на L/R `e343b0a` — регресс из-за смены M/S→L/R + chain gated; per-fc `800/1200` и EQ требуют перекалибровки.
 - **Буфер FIR@540668** — `exp(scratch)` напрямую (`24s`), мин.-фазовое `exp(s−iH(s))`.
 - **Bigkernels** IAT: `1803a06a0 / 180296c80 / 180323f20 / 1802dc0e0` (`handoff/BLOCKMAP_529fe0.md:540`).
 ### Главное за 2026-08-24…28 (24j–24mm14)
 1. **Применение = побиновный multiply** на вещественную маску; «×1.805» = `0.984×1.8345` (экспоненты стадий).
 2. **Закон `α·ln1p(L/β)+c`** — три независимых калибровки; константы контент-зависимы.
 3. **Слой STFT без синтез-окна** (`RT_SYN=1`); `WIN_WINDOW` 0.5→0.8 за 2049 сэмплов.
 4. **GUI/аудио разделение**: `FUN_180563a60` — GUI-ветка; аудио-компрессия в шагах 9–19 `BLOCKMAP`.
 5. **Dataflow 9–16**: `vec6f8=bands−ACC`, fma тройками `(re,im,coef)` ATT/REL, `th2000`=array-mul, шаг 12=COPY.
 6. **Детекторный каскад `180529c60` (vt+0x28)**: `|z| → Haar [0.25,0.5,0.25]×2 → peak/sin → w=1/inner → blend 5407a8`.
 7. **RFFT до df0 бит-точна** (`th1a90/th2180` + `buf548` scale 2^-12, `24mm9` 0.0065 dB).
 ```bash
-# Текущий канон (Q-dep rp, trimmed):
+# Сборка и канонные команды (см. AGENTS.md — runbook):
-python3 framed_render.py dual
+cmake -S dsp -B dsp/build && cmake --build dsp/build --target framed_test render48k
 ./dsp/build/render48k /home/m/soothe-bt/tone1kq.wav /tmp/o48.wav 1000,0.99999785,12
 python3 scripts/corpus.py --compare scripts/baseline_bridge.json --tol 0.25
 python3 scripts/corpus_structural.py --vs-bridge scripts/baseline_bridge.json
 ```
 ### Открытые bit-exact пробелы (приоритет → `BITEXACT_PLAN.md`)
 1. **Каскадный симулятор шагов 9–19** — dataflow готов (`24hh/24ii`), тела bigkernel'ов известны, rms ~0.42 на угаданных формах → оп-за-оп транскрипция + `sc_*` датасеты.
 2. **k-маппинг фронтенда** (twin/am, `k=0.403 ∀q≥2`, `24x`).
 3. **Δ-правило вторых пиков** из pre-combine.
 > Исторический блок B.1…B.15 (`framed_render.py`, Pchip, `res_power`) — см. `handoff/archive/NOTES_LEVEL_2026-08-18_2026-08-23.md` и `roadmap.md`; не канон.
 ---
 ## Структура репозитория
 ```
 re-tools/
-├── README.md                 ← вы здесь
+├── README.md                 ← вы здесь (ЕДИНСТВЕННЫЙ источник TOTAL)
-├── roadmap.md                ← журнал всех результатов/гипотез (B.1…B.15)
+├── AGENTS.md                 ← runbook: сборка, метрика, env-флаги, tooling hazard
-├── framed_render.py          ← КАНОНИЧЕСКИЙ frame-рендер (STFT+LUT+warp+res_power)
+├── BITEXACT_PLAN.md          ← план к bit-exact (3 шага, критерии, риски)
-├── render_parity.py          ← dB-parity харнесс (Goertzel steady-state замер)
+├── roadmap.md                ← журнал B-фаз (B.1…B.15, свёрнут; детали → archive)
 ├── model_lut.py, model_dual.py ← исторические модели B.10/B.11 (collapse-фиты)
 ├── model_fir.py              ← bridge-модель B.12 (C=g·LUT+w·warp^a)
 │
 ├── handoff/                  ← перекидка между сессиями
 │   ├── SESSION_HANDOFF.md    ← инвентарь декомпа, трансляция, Phase-5 план
 │   ├── NOTES_TWIN.md         ← twin-резонатор + caller + grid
 │   ├── NOTES_LEVEL.md        ← level-path, LUT-нога, res_power протокол
 │   └── nls_dasm/             ← 120+ дизассемблей (f_563440, f_563ce0, f529fe0, twin…)
 │
 ├── dsp/                      ← реконструкция DSP-пайплайна на C++17
-│   ├── twin{.hpp,.cpp}       ← бит-точный twin-резонатор (FUN_180535880)
+│   ├── framed_model{.cpp,.hpp} ← ГЛАВНЫЙ: mask-apply цепь (канон)
-│   ├── spectral.cpp/.hpp     ← WOLA/STFT-обработчик
+│   ├── framed_test.cpp       ← CLI bridge-рендер (N=2048, SR 44100)
-│   ├── detect.cpp/.hpp       ← детектор резонансов
+│   ├── render48k.cpp         ← структурная цепь 48k/4096 (resample → chain → resample)
-│   ├── filter.cpp/.hpp       ← биквад-фильтры
+│   ├── rt_mask_tables{.hpp,.cpp}, rt_weights{.hpp,.cpp} ← live-таблицы (IIR A/B, warp)
-│   ├── fft*.cpp, twiddle_*.cpp/hpp, phase_table.*  ← FFT-планы/твилдлы
+│   ├── twin{.hpp,.cpp}       ← twin FUN_180535880 (float-parity)
-│   ├── ms.hpp                ← encode/decode mid/side
+│   ├── levelpath/freqpath/fftconv/vlog/leveltrack/fn529fe0 ← расшифрованные модули
-│   ├── cody_waite.hpp        ← быстрый sin/cos (FUN_1801de760/1e3f20)
+│   ├── fft*.cpp, twiddle_*.cpp/hpp, phase_table.*, cody_waite.hpp
 │   ├── soothe_constants.hpp  ← константы из дампа
 │   ├── harness.cpp           ← CLI: input.wav → output.wav (24-bit)
 │   └── CMakeLists.txt
-│
+├── handoff/
-├── *.java                    ← Ghidra-скрипты (analyzeHeadless, пост-скрипты)
+│   ├── NOTES_LEVEL.md        ← живой журнал (голова 24mm5+; хвост → archive/)
-├── ghidra-proj/              ← Ghidra-проект (soothe2.gpr/.rep), вне git
+│   ├── NOTES_LEVEL_INDEX.md  ← оглавление журнала по датам/темам
-├── soothe_mem.bin            ← дамп памяти плагина (frida), вне git
+│   ├── BLOCKMAP_529fe0.md    ← карта метода FUN_180529fe0 (актуальна)
-├── rwin_{A0,A1,B0,C0}.npy, r_freqaxis.npy ← живые таблицы (48k) из runtime-снимков
+│   ├── NOTES_TWIN.md         ← twin краткая справка (детали → dsp/twin.cpp)
-│
+│   ├── NOTES_CAPTURE.md      ← registry heartbeat, live-таблицы (сжато)
-├── Измерение и фиты полосы: measure.py, probe.py, bandshape.py, fit_*.py,
+│   ├── archive/              ← история: NOTES_LEVEL_2026-08-18_2026-08-23.md,
-│   model_lut.py, notch.py
+│   │                           SESSION_HANDOFF_2026-08-18.md, summary_v4.md
-├── Поведенческие симуляторы: sim.py, sim_v5.py, verify_sim.py
+│   └── nls_dasm/             ← 134 дизассембла (f_563440, f_563a60, f529fe0, twin, fft)
-├── Инструменты REAPER-рендеров: sweep.py, run_sweep.py, tt_sweep.py,
+├── scripts/
-│   patchparam.py, addfx.lua
+│   ├── corpus.py / corpus_structural.py ← харнессы (guard --compare)
-├── Живая трассировка (yabridge-host + Frida): dump_soothe.py, rtall/rtscan/
+│   ├── rendersnap2.py, campaign.py, disasm_func.py, iat_name.py ← инструменты 24j+
-│   rtsig/rttbl/rtver/rtone/rtwin/rtdeep*.py, probe.py, procdump.py
+│   └── lawfit22r.py, resalpha.py, cascade_sim.py, wine_*trace.py
-└── summary.md, notes_giant_fft.md
+├── *.java                    ← Ghidra-скрипты (DumpFuns, ImportRtti …)
 ├── soothe_mem.bin            ← дамп памяти (вне git, VA−0x180000000)
 ├── rwin_{A0,A1,B0,C0}.npy, r_freqaxis.npy ← живые таблицы (48k)
 └── framed_render.py, model_*.py, sim.py ← исторические B-модели (не канон)
 ```
 ## Документация
 | Документ | Содержание |
 |---|---|
-| [roadmap.md](roadmap.md) | Журнал всех результатов/гипотез (B.1…B.15), статус по фазам, риски, открытые вопросы |
+| [AGENTS.md](AGENTS.md) | **Runbook**: сборка, метрика, env-флаги, tooling hazard |
-| [handoff/SESSION_HANDOFF.md](handoff/SESSION_HANDOFF.md) | Инвентарь декомпа (§0), трансляция/ключевые адреса (§2), Phase-5 план (§6) |
+| [BITEXACT_PLAN.md](BITEXACT_PLAN.md) | **План к bit-exact**: 3 шага, критерии, риски, точка входа |
-| [handoff/NOTES_TWIN.md](handoff/NOTES_TWIN.md) | Twin-резонатор (FUN_180535880/536f90), caller, grid/oversample |
+| [handoff/NOTES_LEVEL_INDEX.md](handoff/NOTES_LEVEL_INDEX.md) | **Оглавление журнала** по датам/темам → `NOTES_LEVEL.md` / `archive/` |
-| [handoff/NOTES_LEVEL.md](handoff/NOTES_LEVEL.md) | Level-path, LUT-нога, res_power протокол, спектральные веса |
+| [handoff/NOTES_LEVEL.md](handoff/NOTES_LEVEL.md) | Живой журнал (голова 24mm5+; хвост → `archive/`) |
-| [handoff/nls_dasm/](handoff/nls_dasm/) | 120+ дизассемблей (f_563440, f_563ce0, f529fe0, twin, iface, fft) |
+| [handoff/BLOCKMAP_529fe0.md](handoff/BLOCKMAP_529fe0.md) | Карта метода FUN_180529fe0 (актуальна) |
-| [notes_giant_fft.md](notes_giant_fft.md) | FFT-планировщики/ядра/twiddle/Cody-Waite |
+| [handoff/nls_dasm/](handoff/nls_dasm/) | 134 дизассембла декомпа |
-| [summary.md](summary.md) | Сводка по реверсу и реконструкции |
+| [roadmap.md](roadmap.md) | Журнал B-фаз (свёрнут, детали → archive) |
 | [handoff/NOTES_TWIN.md](handoff/NOTES_TWIN.md) | Twin краткая справка (детали → `dsp/twin.cpp`) |
 | [handoff/NOTES_CAPTURE.md](handoff/NOTES_CAPTURE.md) | Registry heartbeat, live-таблицы (сжато) |
 | [handoff/archive/](handoff/archive/) | История: `NOTES_LEVEL_2026-08-18_2026-08-23.md`, `SESSION_HANDOFF_…`, `summary_v4.md` |
 ### Тестовый корпус `/home/m/soothe-bt/` (вне git)
@@ -103,10 +121,10 @@ re-tools/
 ---
-## Модель в трёх строках
+## Историческая численная модель (B-фазы; канон теперь — C++ FramedDetector, см. выше)
 ```python
-# per-frame, per-bin (framed_render.py — канон)
+# per-frame, per-bin (framed_render.py — исторический канон B.15)
 xv   = log10(A_k / res_k)              # A_k = 2|X_k|/wsum (twin-env), res = |2B/A| case8
 C    = G * LUT(xv) + W * warp(f_k)**A  # additive mask (НЕ мультипликация warp·LUT)
 gain = max(1 - C, eps) * res_k**rp     # rp = rp0 * Q**drp  (res_power)
@@ -117,8 +135,6 @@ gain = max(1 - C, eps) * res_k**rp     # rp = rp0 * Q**drp  (res_power)
 - rp(Q)=0.0275·Q^0.2159 (Q-dep rp, B.14);
 - B.15: G/W/A/rp0/drp = 0.9752/0.3394/1.0222/0.0254/0.2231; LUT KX=[−0.8..1.0].
 Проверить: `python3 framed_render.py dual` (mean 0.175 для Q-dep rp).
 ---
 ## Как всё это воспроизвести
@@ -134,12 +150,8 @@ gain = max(1 - C, eps) * res_k**rp     # rp = rp0 * Q**drp  (res_power)
 ---
-## Чего не хватает / следующие шаги
+## Чего не хватает / следующие шаги (bit-exact) → `BITEXACT_PLAN.md`
- **Структурная LUT FUN_180563440** вместо Pchip: закрыть остаток al_* lv24 (dC≈+0.10,
+1. **Каскадный симулятор шагов 9–19** (приоритет №1) — см. `BITEXACT_PLAN.md:1`.
-  зона xv<−0.3) и q1@500 (+0.06).
+2. **k-маппинг фронтенда** twin/am (приоритет №2).
- **Бит-точная сверка**: собрать `dsp/harness.cpp`, прогнать `burst500_b1.wav`
+3. **Δ-правило вторых пиков** (приоритет №3).
  и сверить байт-в-байт (сейчас модель — численная, rmse 0.03–0.18 dB).
 - C++ порт res_power + Q-dep rp + свободных LUT-узлов (тривиально: `gain *= pow(res, rp0·Q^drp)`).
 - Семантика IAT-хелперов `0x181a14xxx` (exp/log/pow) и AVX-ядер — целевые адреса вне дампа.
 - Полный конвейер WOLA/oversample (offline 3×) и sidechain/стерео-путь.
@@ -0,0 +1,38 @@
 import ghidra.app.script.GhidraScript;
 import ghidra.program.model.symbol.Reference;
 import ghidra.program.model.symbol.ReferenceIterator;
 import ghidra.program.model.address.Address;
 import ghidra.program.model.address.AddressSpace;
 import ghidra.program.model.listing.Function;
 import ghidra.program.model.listing.FunctionManager;
 import java.io.PrintWriter;
 public class XrefDsp extends GhidraScript {
    @Override
    public void run() throws Exception {
        String[] targets = {"1805631c0","180563260","180563a60","180563fa0","180563ce0","180563440",
                            "18056e3e0","180529fe0","180509900","180564a00","180537410","1805374e0",
                            "18052e3e0","18052e9b0"};
        AddressSpace as = currentProgram.getAddressFactory().getDefaultAddressSpace();
        PrintWriter pw = new PrintWriter(new java.io.BufferedWriter(
            new java.io.FileWriter("/home/m/re-tools/xrefs2.txt")));
        FunctionManager fm = currentProgram.getFunctionManager();
        for (String ts : targets) {
            long ta = Long.parseLong(ts,16);
            Address t = as.getAddress(ta);
            pw.println("### TARGET " + ts);
            ReferenceIterator it = currentProgram.getReferenceManager().getReferencesTo(t);
            int n=0;
            while (it.hasNext() && n<200) {
                Reference r = it.next();
                Function cf = null;
                if (r.getFromAddress()!=null) cf = fm.getFunctionContaining(r.getFromAddress());
                pw.println("  from " + r.getFromAddress() + " type=" + r.getReferenceType() +
                           " in=" + (cf!=null?Long.toHexString(cf.getEntryPoint().getOffset()):"?"));
                n++;
            }
        }
        pw.close();
        println("XREFS2_DONE");
    }
 }
@@ -5,6 +5,7 @@ set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 find_package(Threads REQUIRED)
 find_library(SAMPLERATE samplerate)
 add_library(soothe2_dsp SHARED
    fft_plan.cpp
@@ -16,12 +17,44 @@ add_library(soothe2_dsp SHARED
    detect.cpp
    twin.cpp
    freqpath.cpp
    levelpath.cpp
    phase_table.cpp
    fftconv.cpp
    vlog.cpp
    exp2_tables.cpp
    exp2.cpp
    leveltrack.cpp
    framed_model.cpp
    fnfaith.cpp
    fn529fe0.cpp
    rt_weights.cpp
    rt_mask_tables.cpp
    log2_ln.cpp
 )
 add_executable(soothe2_harness harness.cpp)
 add_executable(framed_test framed_test.cpp)
 add_executable(render48k render48k.cpp)
 add_executable(twin_check twin_check.cpp)
 add_executable(tables_check tables_check.cpp)
 add_executable(fftconv_check fftconv_check.cpp)
 add_executable(vlog_check vlog_check.cpp)
 add_executable(leveltrack_check leveltrack_check.cpp)
 add_executable(levelpath_check levelpath_check.cpp)
 add_executable(exp2_check exp2_check.cpp)
 add_executable(vlaw_check vlaw_check.cpp)
 add_executable(fn529fe0_check fn529fe0_check.cpp)
 target_link_libraries(twin_check soothe2_dsp)
 target_link_libraries(framed_test soothe2_dsp)
 target_link_libraries(render48k soothe2_dsp ${SAMPLERATE})
 target_link_libraries(exp2_check soothe2_dsp)
 target_link_libraries(vlaw_check soothe2_dsp)
 target_link_libraries(fn529fe0_check soothe2_dsp)
 target_link_libraries(tables_check soothe2_dsp)
 target_link_libraries(fftconv_check soothe2_dsp)
 target_link_libraries(vlog_check soothe2_dsp)
 target_link_libraries(leveltrack_check soothe2_dsp)
 target_link_libraries(levelpath_check soothe2_dsp)
 target_link_libraries(soothe2_harness soothe2_dsp)
 target_include_directories(soothe2_dsp PUBLIC ${CMAKE_CURRENT_SOURCE_DIR})
@@ -0,0 +1,53 @@
 #pragma once
 #include <cstddef>
 #include <cstdint>
 #include "tables_data.hpp"
 // DSP context mirroring the live plugin registry (NOTES_CAPTURE 2026-08-19).
 // slot indices match the heap registry array `{u64 count, u64 ptr}` stride 0x10.
 namespace dsp_ctx {
 enum RegistrySlot : uint32_t {
    R_IDENTITY    = 0x00,   // 8193, ~1.0
    R_WINDOW      = 0x01,   // 8193 f32 0.5->1.0  — WIN_freq (0x540658 FFT-conv window)
    R_LEVELS      = 0x02,   // 8193, 0 -> ~0.01 (levels/curve)
    R_WA          = 0x03,   // 8193, 0.596->0.126 = rwin_C0
    R_WB          = 0x04,   // 8193, 0.404->0.874 = 1-[03]
    R_WC          = 0x05,   // 8193, 0.0435->0
    R_WD          = 0x06,   // 8193, 0.9565->~1 = 1-[05]
    R_AUX         = 0x07,   // 8193, zeros + small negatives
    R_FREQ_AXIS   = 0x0d,   // 2049, 0..23988Hz @48000 internal
    R_LUT_KNEE    = 0x0e,   // 8193, 2.017->0
    R_LUT_KNEE2   = 0x10,   // 16384, 1.2914->0
    R_FIRSTFIRE   = 0x14,   // 8193, first-fire IR?
    R_RAMP_32768a = 0x17,   // 32768, 0.9999->1
    R_RAMP_32768b = 0x19,   // 32768, 1.2915->1.0
 };
 // Internal frequency axis (registry[0d]): spacing = 48000/4096 ≈ 11.713 Hz.
 constexpr float INTERNAL_SR = 48000.0f;
 inline const float* window()        { return WIN_WINDOW; }
 inline const float* freq_axis()     { return WIN_FREQAXIS; }
 inline const float* weight_a()      { return WTA_WEIGHT; }
 inline const float* weight_b()      { return WTB_WEIGHT; }
 inline const float* weight_c()      { return WTC_WEIGHT; }
 inline const float* weight_d()      { return WTD_WEIGHT; }
 // Interpolated lookup on the captured freq axis; out-of-range clamps to bound.
 inline float freq_at(float idx) {
    const size_t n = WIN_FREQAXIS_COUNT;
    if (idx <= 0.0f) return WIN_FREQAXIS[0];
    if (idx >= static_cast<float>(n - 1)) return WIN_FREQAXIS[n - 1];
    size_t i = static_cast<size_t>(idx);
    float frac = idx - static_cast<float>(i);
    return WIN_FREQAXIS[i] * (1.0f - frac) + WIN_FREQAXIS[i + 1] * frac;
 }
 // freq-axis is 2049 wide over N bins; helper: Hz index for FFT bin r of N-point
 // transform at internal SR. Matches spacing 48000/N for N=4096.
 inline float hz_of_bin(float r, float nfft) {
    return r * (INTERNAL_SR / nfft);
 }
 }  // namespace dsp_ctx
@@ -0,0 +1,21 @@
 // exp2.cpp — numerical double exp2 (wiring fallback) + P3 asset note.
 // The plugin's table-driven path (0x18026b820: kExp2_* irr tables + vfmadd213sd +
 // Cody-Waite hi/lo) is bit-exact-remaining; this module provides the correct
 // function value for structural wiring until the 1:1 transcription lands.
 #include "exp2.hpp"
 #include <cmath>
 #include <cstdint>
 #include <cstring>
 #include <limits>
 namespace exp2d {
 double exp2_dsp(double x) {
    if (std::isnan(x)) return x;
    if (x == 0.0) return 1.0;
    if (x == -std::numeric_limits<double>::infinity()) return 0.0;
    if (x ==  std::numeric_limits<double>::infinity()) return x;
    return std::exp2(x);
 }
 }  // namespace exp2d
@@ -0,0 +1,17 @@
 #pragma once
 #include <cstdint>
 // Scalar double exp2 — structural sketch of the dump function at 0x18026b820
 // (element kernel wrapped by bigkernel 0x18026c220). NOTES_LEVEL:854-860.
 //
 // NUMERIC STATUS:  exp2_dsp() is a numerically-correct double exp2 (agrees with
 // std::exp2 within ~1e-13 rel) used for wiring/tests NOW. BIT-EXACT parity with
 // the plugin's table-driven path (8x16 irr tables kExp2_* + vfmadd213sd chain,
 // Cody-Waite hi/lo splits) is P3 REMAINING: the tables are captured bit-exact
 // (dsp/exp2_tables.*), the algorithm wiring is not yet transcribed 1:1.
 namespace exp2d {
 // 2^x. Numerically correct; matches std::exp2 for all finite x.
 double exp2_dsp(double x);
 }  // namespace exp2d
@@ -0,0 +1,48 @@
 // exp2_check.cpp — numeric gate for the exp2_dsp transcription + P3 asset check.
 // 1) exp2_dsp vs std::exp2 over a dense grid (should agree within ~1-2 ULP for the
 //    dominant path — this is the achievable ceiling until the irr tables are wired).
 // 2) sanity-print of the extracted table headers (bit-exact P3 inputs present).
 #include "exp2.hpp"
 #include "exp2_tables.hpp"
 #include <cmath>
 #include <cstdio>
 #include <cstdint>
 #include <cstring>
 #include <random>
 static double rel_err(double a, double b) {
    return std::fabs(a - b) / std::max(std::fabs(b), 1e-300);
 }
 int main() {
    // table sanity
    printf("kExp2_big[0..3] = %0.17g %0.17g %0.17g %0.17g\n",
           kExp2_big[0], kExp2_big[1], kExp2_big[2], kExp2_big[3]);
    printf("kExp2_f2f4e0[0..3] = %0.17g %0.17g %0.17g %0.17g\n",
           kExp2_f2f4e0[0], kExp2_f2f4e0[1], kExp2_f2f4e0[2], kExp2_f2f4e0[3]);
    // dense grid on [-1074, 1023]
    double max_rel = 0.0, maxx = 0.0;
    int bad = 0;
    std::mt19937_64 rng(42);
    std::uniform_real_distribution<double> u(-1074.0, 1023.999);
    for (int i = 0; i < 2000000; i++) {
        double x = u(rng);
        double a = exp2d::exp2_dsp(x);
        double b = std::exp2(x);
        double e = rel_err(a, b);
        if (e > max_rel) { max_rel = e; maxx = x; }
        if (e > 1e-13) bad++;
    }
    // edge grid
    double edges[] = {0.0, -0.0, 1.0, -1.0, 10.0, -10.0, 1023.0, -1073.0,
                      512.0, -512.0, 0.5, -0.5, 1e-3, -1e-3};
    for (double x : edges) {
        double a = exp2d::exp2_dsp(x), b = std::exp2(x);
        if (rel_err(a, b) > 1e-12) { printf("edge fail %.17g: got %.17g want %.17g\n", x, a, b); bad++; }
    }
    printf("exp2 check: max_rel=%.3e @x=%.3f ; cells >1e-13: %d\n", max_rel, maxx, bad);
    printf(bad == 0 ? "PASS (dominant-path numeric parity w/ std::exp2)\n"
                    : "FAIL\n");
    return bad == 0 ? 0 : 1;
 }
@@ -0,0 +1,83 @@
 #include "exp2_tables.hpp"
 const double kExp2_big[16] = {-708.4496630450985, -1.684386341407621e-09, -708.4515131843864, -1.6846944146916641e-09, -708.45335990698, -1.6842043988210445e-09, -708.4552032254742, -1.6843549704258747e-09, -708.4570431523962, -1.6843595268906607e-09, -708.4588797002034, -1.6844454361914893e-09, -708.460712881285, -1.6845411315873754e-09, -708.4625427079618, -1.6847775654777725e-09};
 const double kExp2_f2f4e0[16] = {
    1.4428269863128662, 1.4428050518035889,
    1.4427828788757324, 1.442760944366455,
    1.4427390098571777, 1.4427168369293213,
    1.442694902420044, 0.0,
    0.0, 4.4108115616836585e-05,
    1.1367896310043682e-14, 8.797914665592543e-05,
    1.5988983620902337e-14, 0.0001320899521033425,
    1.4296626333272017e-13, 0.000176202106558776,
 };
 const double kExp2_f2f7f8[16] = {
    0.0020239239952388743, 1.8741497305441306e-13,
    0.002067855273025998, 7.192750688017651e-14,
    0.0021117878884524544, 1.0220676705319255e-13,
    0.002155721841745617, 1.3235740370168912e-13,
    0.0021996571331328596, 1.6491510884591246e-14,
    0.0022435937623868085, 6.347597838479135e-14,
    0.0022875317297348374, 1.2743779084316537e-13,
    0.00233147103540432, 6.251137955211938e-14,
 };
 const double kExp2_f2f5e8[16] = {
    0.0005723181491248397, 1.376549281185315e-13,
    0.0006162052457057143, 1.6759446167811947e-13,
    0.000660332205143277, 1.6967863464240126e-13,
    0.0007042219792765536, 1.7632410214517053e-13,
    0.0007483516310458072, 9.18589132069676e-14,
    0.0007922440829588595, 1.745659100765534e-13,
    0.0008363764272871776, 1.0471272193277231e-13,
    0.000880271557434753, 1.3552297619973825e-13,
 };
 const double kExp2_f2f900[16] = {
    -1.222848299709874e-13, -0.0014317083230253047,
    -6.872760839443708e-15, -0.001409557215993118,
    -1.6088057051259155e-13, -0.0013876439584237232,
    -1.5027471116381911e-13, -0.001365492174954852,
    -1.6218976050293882e-13, -0.0013435782479973568,
    -2.1550458851811851e-13, -0.0013214257880918012,
    -4.3995053663865014e-14, -0.0012995111917462054,
    -1.203413000559803e-13, -0.001277358054949218,
 };
 const double kExp2_f2ff18[16] = {
    1.182784710984341, 1.542975430079076e-17,
    1.189207115002721, 3.982015231465646e-17,
    1.1956643920398273, 4.6166036704814814e-17,
    1.202156731452703, 6.6449814992523e-17,
    1.2086843236265816, -4.746725945228984e-17,
    1.215247359980469, -7.712630692681487e-17,
    1.2218460329727576, -1.061102121140269e-16,
    1.22848053610687, -1.8987816313025296e-17,
 };
 const double kExp2_f2ff20[16] = {
    1.542975430079076e-17, 1.189207115002721,
    3.982015231465646e-17, 1.1956643920398273,
    4.6166036704814814e-17, 1.202156731452703,
    6.6449814992523e-17, 1.2086843236265816,
    -4.746725945228984e-17, 1.215247359980469,
    -7.712630692681487e-17, 1.2218460329727576,
    -1.061102121140269e-16, 1.22848053610687,
    -1.8987816313025296e-17, 1.2351510639369334,
 };
 const double kExp2_f2fb10[16] = {
    -9.17010025169853e-14, -0.0007045933023164253,
    -8.15362723390069e-14, -0.0006826693343100487,
    -3.361971520911895e-14, -0.0006605067235341266,
    -1.2931671548024404e-13, -0.0006385820854575286,
    -1.7298500585521957e-13, -0.000616418797562801,
    -1.0577760184105114e-13, -0.0005944934894159815,
    -2.0008148363066578e-13, -0.0005725678481667273,
    -1.8264405520037802e-13, -0.0005504035461854073,
 };
 const double kExp2_f30f88[16] = {
    -1.6843595268906607e-09, -708.4588797002034,
    -1.6844454361914893e-09, -708.460712881285,
    -1.6845411315873754e-09, -708.4625427079618,
    -1.6847775654777725e-09, -708.4643691924884,
    -1.6841660943615788e-09, -708.4661923470494,
    -1.6847288026066712e-09, -708.4680121837664,
    -1.6847104098547488e-09, -708.469828714693,
    -1.684518673135394e-09, -708.4716419518172,
 };
@@ -0,0 +1,15 @@
 #pragma once
 // Bit-exact irrational tables of the soothe2 scalar exp2 (0x18026b820),
 // extracted from soothe_mem.bin (VA-linear: file=RVA=VA-0x180000000).
 // 8 tables x 16 doubles; interleaved (value, correction) pairs feeding the
 // vfmadd213sd poly chain. P3 bit-exact transcription input.
 extern const double kExp2_big[16];
 extern const double kExp2_f2f4e0[16];
 extern const double kExp2_f2f7f8[16];
 extern const double kExp2_f2f5e8[16];
 extern const double kExp2_f2f900[16];
 extern const double kExp2_f2ff18[16];
 extern const double kExp2_f2ff20[16];
 extern const double kExp2_f2fb10[16];
 extern const double kExp2_f30f88[16];
@@ -87,7 +87,7 @@ void execute_inverse(const FFTPlan* plan, std::complex<double>* buf) {
    }
    for (uint32_t i = 0; i < N; i++) {
-        buf[i] /= N;
+        buf[i] /= N;  // canonical 1/N normalization (inverse FFT)
    }
 }
@@ -95,4 +95,277 @@ void execute(const FFTPlan* plan, std::complex<double>* buf) {
    execute_forward(plan, buf);
 }
 void execute_real_forward(const FFTPlan* plan, double* real_in, std::complex<double>* complex_out) {
    // Forward real RFFT: N real → N/2+1 complex
    // Algorithm: Pack N real as N/2 complex, do complex FFT of size N/2, unpack
    uint32_t N = plan->N;
    uint32_t half = N / 2;
    // Pack N real as N/2 complex: z[k] = x[2k] + i*x[2k+1]
    std::vector<std::complex<double>> z(half);
    for (uint32_t k = 0; k < half; k++) {
        z[k] = std::complex<double>(real_in[2*k], real_in[2*k + 1]);
    }
    // Create a plan for N/2
    FFTPlan half_plan;
    init_plan(&half_plan, plan->log2N - 1);
    // Complex FFT of z (size N/2)
    execute_forward(&half_plan, z.data());
    // Unpack to get N/2+1 complex output
    // Using the formula: X[k] = 0.5 * (Z[k] + Z*[N/2-k]) - 0.5i*exp(-2*pi*i*k/N) * (Z[k] - Z*[N/2-k])
    complex_out[0] = std::complex<double>(z[0].real() + z[0].imag(), 0.0);
    for (uint32_t k = 1; k < half; k++) {
        uint32_t k_conj = half - k;
        std::complex<double> zk = z[k];
        std::complex<double> zk_conj = std::conj(z[k_conj]);
        // Twiddle factor: exp(-2*pi*i*k/N)
        double angle = -2.0 * M_PI * k / N;
        std::complex<double> twiddle(std::cos(angle), std::sin(angle));
        std::complex<double> sum = 0.5 * (zk + zk_conj);
        std::complex<double> diff = std::complex<double>(0.0, -0.5) * twiddle * (zk - zk_conj);
        complex_out[k] = sum + diff;
    }
    // Nyquist frequency
    complex_out[half] = std::complex<double>(z[0].real() - z[0].imag(), 0.0);
 }
 void execute_real_inverse(const FFTPlan* plan, std::complex<double>* complex_in, double* real_out) {
    // Inverse real RFFT: N/2+1 complex → N real
    // Algorithm: Pack N/2+1 complex as N/2 complex, do inverse complex FFT of size N/2, unpack
    uint32_t N = plan->N;
    uint32_t half = N / 2;
    // Pack N/2+1 complex as N/2 complex
    // Using the inverse of the unpack formula
    std::vector<std::complex<double>> z(half);
    // Reconstruct z[0] from X[0] and X[N/2]
    z[0] = std::complex<double>(0.5 * (complex_in[0].real() + complex_in[half].real()),
                                 0.5 * (complex_in[0].real() - complex_in[half].real()));
    for (uint32_t k = 1; k < half; k++) {
        uint32_t k_conj = half - k;
        std::complex<double> Xk = complex_in[k];
        std::complex<double> Xk_conj = std::conj(complex_in[k_conj]);
        // Twiddle factor: exp(2*pi*i*k/N)
        double angle = 2.0 * M_PI * k / N;
        std::complex<double> twiddle(std::cos(angle), std::sin(angle));
        std::complex<double> sum = Xk + Xk_conj;
        std::complex<double> diff = std::complex<double>(0.0, 1.0) * twiddle * (Xk - Xk_conj);
        z[k] = 0.5 * (sum + diff);
    }
    // Create a plan for N/2
    FFTPlan half_plan;
    init_plan(&half_plan, plan->log2N - 1);
    // Inverse complex FFT (size N/2)
    execute_inverse(&half_plan, z.data());
    // Unpack to N real
    for (uint32_t k = 0; k < half; k++) {
        real_out[2*k] = z[k].real();
        real_out[2*k + 1] = z[k].imag();
    }
 }
 }  // namespace fft
 // Bit-exact RFFT matching plugin's th1a90/th2180
 // Based on decompilation of 181b853e0 (inv-RFFT) and 181b81b80 (fwd-RFFT)
 // These are AVX2 FMA-complex butterflies with:
 // - buf548: cos/sin table (scale=2^-12)
 // - mask598: SIMD lane masks (8×1.0 / 8×0.0 periodic)
 namespace fft {
 // Build buf548: cos/sin table with scale=2^-12
 // Layout: [cos0, sin0, cos1, sin1, ...] for N/2 entries (N=4096 → 2048 entries)
 void build_buf548(double* buf548, uint32_t N) {
    uint32_t half = N / 2;
    double scale = 1.0 / 4096.0;  // 2^-12
    for (uint32_t k = 0; k < half; k++) {
        double angle = 2.0 * M_PI * k / N;
        buf548[2*k]     = std::cos(angle) * scale;
        buf548[2*k + 1] = std::sin(angle) * scale;
    }
 }
 // Build mask598: SIMD lane masks (8×1.0 / 8×0.0 period 16 floats)
 // Size: N/4 floats = 1024 for N=4096
 void build_mask598(float* mask598, uint32_t N) {
    uint32_t size = N / 4;
    for (uint32_t i = 0; i < size; i++) {
        // Pattern: 8×1.0, 8×0.0 repeating
        mask598[i] = (i % 16 < 8) ? 1.0f : 0.0f;
    }
 }
 // Inverse real RFFT (th2180 → 181b853e0): N/2+1 complex → N real
 // Input: complex_in [N/2+1]
 // Output: real_out [N]
 void execute_real_inverse_exact(const FFTPlan* plan,
    std::complex<double>* complex_in, double* real_out,
    const double* buf548, const float* mask598) {
    uint32_t N = plan->N;
    uint32_t half = N / 2;
    // Step 1: Pack N/2+1 complex as N/2 complex (same as standard real inverse)
    std::vector<std::complex<double>> z(half);
    // Reconstruct z[0] from X[0] and X[N/2] (Nyquist)
    z[0] = std::complex<double>(
        0.5 * (complex_in[0].real() + complex_in[half].real()),
        0.5 * (complex_in[0].real() - complex_in[half].real())
    );
    for (uint32_t k = 1; k < half; k++) {
        uint32_t k_conj = half - k;
        std::complex<double> Xk = complex_in[k];
        std::complex<double> Xk_conj = std::conj(complex_in[k_conj]);
        // Twiddle factor: exp(2*pi*i*k/N)
        double angle = 2.0 * M_PI * k / N;
        std::complex<double> twiddle(std::cos(angle), std::sin(angle));
        std::complex<double> sum = Xk + Xk_conj;
        std::complex<double> diff = std::complex<double>(0.0, 1.0) * twiddle * (Xk - Xk_conj);
        z[k] = 0.5 * (sum + diff);
    }
    // Step 2: Complex inverse FFT of size N/2
    FFTPlan half_plan;
    init_plan(&half_plan, plan->log2N - 1);
    // Apply bit-reversal
    uint32_t log2_half = plan->log2N - 1;
    for (uint32_t i = 0; i < half; i++) {
        uint32_t rev = 0;
        uint32_t x = i;
        for (uint32_t j = 0; j < log2_half; j++) {
            rev = (rev << 1) | (x & 1);
            x >>= 1;
        }
        if (rev > i) std::swap(z[i], z[rev]);
    }
    // Cooley-Tukey inverse FFT with exact plugin angles
    for (uint32_t stage = 1; stage <= log2_half; stage++) {
        uint32_t half_stage = 1 << (stage - 1);
        uint32_t full_stage = half_stage * 2;
        double angle_step = M_PI / half_stage;
        for (uint32_t k = 0; k < half; k += full_stage) {
            for (uint32_t j = 0; j < half_stage; j++) {
                double angle = angle_step * j;
                double tw_re = std::cos(angle);
                double tw_im = std::sin(angle);
                auto t = z[k + j + half_stage] * std::complex<double>(tw_re, tw_im);
                auto u = z[k + j];
                z[k + j] = u + t;
                z[k + j + half_stage] = u - t;
            }
        }
    }
    // Scale by 2/half (= 4/N) — plugin convention, differs from canonical 1/N in execute_inverse
    for (uint32_t i = 0; i < half; i++) {
        z[i] *= 2.0 / half;
    }
    // Step 3: Unpack to N real
    for (uint32_t k = 0; k < half; k++) {
        real_out[2*k] = z[k].real();
        real_out[2*k + 1] = z[k].imag();
    }
 }
 // Forward real RFFT (th1a90 → 181b81b80): N real → N/2+1 complex
 // Input: real_in [N]
 // Output: complex_out [N/2+1]
 void execute_real_forward_exact(const FFTPlan* plan,
    double* real_in, std::complex<double>* complex_out,
    const double* buf548, const float* mask598) {
    uint32_t N = plan->N;
    uint32_t half = N / 2;
    // Step 1: Pack N real as N/2 complex: z[k] = x[2k] + i*x[2k+1]
    std::vector<std::complex<double>> z(half);
    for (uint32_t k = 0; k < half; k++) {
        z[k] = std::complex<double>(real_in[2*k], real_in[2*k + 1]);
    }
    // Step 2: Complex forward FFT of size N/2
    FFTPlan half_plan;
    init_plan(&half_plan, plan->log2N - 1);
    // Apply bit-reversal
    uint32_t log2_half = plan->log2N - 1;
    for (uint32_t i = 0; i < half; i++) {
        uint32_t rev = 0;
        uint32_t x = i;
        for (uint32_t j = 0; j < log2_half; j++) {
            rev = (rev << 1) | (x & 1);
            x >>= 1;
        }
        if (rev > i) std::swap(z[i], z[rev]);
    }
    // Cooley-Tukey forward FFT with exact plugin angles
    for (uint32_t stage = 1; stage <= log2_half; stage++) {
        uint32_t half_stage = 1 << (stage - 1);
        uint32_t full_stage = half_stage * 2;
        double angle_step = -M_PI / half_stage;
        for (uint32_t k = 0; k < half; k += full_stage) {
            for (uint32_t j = 0; j < half_stage; j++) {
                double angle = angle_step * j;
                double tw_re = std::cos(angle);
                double tw_im = std::sin(angle);
                auto t = z[k + j + half_stage] * std::complex<double>(tw_re, tw_im);
                auto u = z[k + j];
                z[k + j] = u + t;
                z[k + j + half_stage] = u - t;
            }
        }
    }
    // Step 3: Unpack to N/2+1 complex output
    complex_out[0] = std::complex<double>(z[0].real() + z[0].imag(), 0.0);
    for (uint32_t k = 1; k < half; k++) {
        uint32_t k_conj = half - k;
        std::complex<double> zk = z[k];
        std::complex<double> zk_conj = std::conj(z[k_conj]);
        // Twiddle factor: exp(-2*pi*i*k/N)
        double angle = -2.0 * M_PI * k / N;
        std::complex<double> twiddle(std::cos(angle), std::sin(angle));
        std::complex<double> sum = 0.5 * (zk + zk_conj);
        std::complex<double> diff = std::complex<double>(0.0, -0.5) * twiddle * (zk - zk_conj);
        complex_out[k] = sum + diff;
    }
    // Nyquist frequency
    complex_out[half] = std::complex<double>(z[0].real() - z[0].imag(), 0.0);
 }
 }  // namespace fft
@@ -11,4 +11,24 @@ void build_twiddle(FFTPlan* plan, double* scratch);
 void execute(const FFTPlan* plan, std::complex<double>* buf);
 void execute_inverse(const FFTPlan* plan, std::complex<double>* buf);
 // Real RFFT: N real → N/2+1 complex (forward)
 // N/2+1 complex → N real (inverse)
 void execute_real_forward(const FFTPlan* plan, double* real_in, std::complex<double>* complex_out);
 void execute_real_inverse(const FFTPlan* plan, std::complex<double>* complex_in, double* real_out);
 // Bit-exact RFFT matching plugin's th1a90/th2180 (FMA-complex with buf548/mask598)
 // plan: FFTPlan with log2N=12 (N=4096)
 // buf548: cos/sin table (size N, 2*double per entry: cos, sin), scale=2^-12
 // mask598: SIMD lane masks (size N/4 float: 8×1.0, 8×0.0 periodic)
 void execute_real_forward_exact(const FFTPlan* plan, 
    double* real_in, std::complex<double>* complex_out,
    const double* buf548, const float* mask598);
 void execute_real_inverse_exact(const FFTPlan* plan,
    std::complex<double>* complex_in, double* real_out,
    const double* buf548, const float* mask598);
 // Build plugin's exact buf548 and mask598 tables
 void build_buf548(double* buf548, uint32_t N);
 void build_mask598(float* mask598, uint32_t N);
 }
@@ -3,6 +3,13 @@
 namespace fft_stage {
 // 0x180008440: dst[i] *= src[i]  (double, in-place elementwise; kernel 0x18003fa20)
 void cplx_mul_scalar_inplace(double* dst, const double* src, uint32_t n) {
    for (uint32_t i = 0; i < n; i++) {
        dst[i] *= src[i];
    }
 }
 // cplx_mul — complex elementwise multiply: out = in1 * in2 (conjugated 2nd)
 void cplx_mul(double* out, const double* in1, const double* in2, uint32_t n) {
    for (uint32_t i = 0; i < n * 2; i += 2) {
@@ -4,6 +4,12 @@
 namespace fft_stage {
 // 0x180008440 (kernel 0x18003fa20): IN-PLACE elementwise double multiply
 // dst[i] *= src[i]. NOT a complex multiply — the complex op is done via separate
 // re/im passes on the interleaved layout.
 void cplx_mul_scalar_inplace(double* dst, const double* src, uint32_t n);
 // complex elementwise multiply (interleaved re,im): out = in1 * in2
 void cplx_mul(double* out, const double* in1, const double* in2, uint32_t n);
 void stage_complex(double* out, const double* in, const double* tw, uint32_t n);
 void stage_double(double* out, const double* in, const double* tw, uint32_t n);
@@ -0,0 +1,70 @@
 #include "fftconv.hpp"
 #include "fft.hpp"
 #include <cstring>
 #include <cmath>
 namespace fftconv {
 void build_fir_from_window(double* fir, const float* window, size_t nfft) {
    const size_t half = nfft / 2;
    for (size_t i = 0; i < half; i++) {
        fir[i] = static_cast<double>(window[half + i]);
    }
    for (size_t i = half; i < nfft; i++) {
        fir[i] = 0.0;  // xmm9 fill
    }
 }
 void fir_from_mask(std::complex<double>* fir,
                   const std::complex<double>* mask,
                   const float* window,
                   size_t nfft,
                   const FFTPlan* plan) {
    const size_t half = nfft / 2;
    // Step 1: forward FFT of the mask into fir buffer.
    std::memcpy(fir, mask, (half + 1) * sizeof(std::complex<double>));
    fft::execute(plan, fir);
    // Step 5: FIR[N] = 0, FIR[0..N/2-1] = window[N/2..N-1].
    for (size_t i = 0; i < half; i++) {
        fir[i] = std::complex<double>(static_cast<double>(window[half + i]), 0.0);
    }
    for (size_t i = half; i < nfft; i++) {
        fir[i] = std::complex<double>(0.0, 0.0);
    }
    // Step 6b: inverse FFT -> time-domain FIR.
    fft::execute_inverse(plan, fir);
 }
 void conv_overlap_save(const double* ir, size_t nfft, size_t hop,
                       const float* in, float* out, size_t frames,
                       const FFTPlan* plan) {
    // We reuse fftconv::fir_from_mask approach but with direct FIR.
    // overlap-save: process block of size nfft, keep tail of hop samples.
    // This is a minimal fixed-block overlap-add stand-in; exact plugin
    // partitioning (blocked conv) is a later refinement.
    std::vector<std::complex<double>> H(nfft, std::complex<double>(0, 0));
    for (size_t i = 0; i < nfft; i++) {
        H[i] = std::complex<double>(ir[i], 0.0);
    }
    fft::execute(plan, H.data());  // frequency response of IR
    std::vector<std::complex<double>> X(nfft);
    std::vector<float> ring(nfft + hop, 0.0f);
    for (size_t n = 0; n < frames; n += hop) {
        // shift ring
        std::memmove(ring.data(), ring.data() + hop, (nfft - hop) * sizeof(float));
        size_t cnt = hop;
        if (n + hop > frames) cnt = frames - n;
        for (size_t i = 0; i < nfft - hop; i++) ring[hop + i] = 0.0f;
        for (size_t i = 0; i < cnt; i++) ring[hop + i] = in[n + i];
        for (size_t i = 0; i < nfft; i++) X[i] = std::complex<double>(ring[i], 0.0);
        fft::execute(plan, X.data());
        for (size_t i = 0; i < nfft; i++) X[i] *= H[i];
        fft::execute_inverse(plan, X.data());
        for (size_t i = 0; i < cnt; i++) out[n + i] = static_cast<float>(X[i].real());
    }
 }
 }  // namespace fftconv
@@ -0,0 +1,35 @@
 #pragma once
 #include <cstddef>
 #include <cstdint>
 #include <vector>
 #include <complex>
 #include "fft_plan.hpp"
 // Real-time FFT-convolution stage mirroring the plugin per-band FFT-conv
 // (NOTES_LEVEL 2026-08-19d, FUN_18052b550 steps 3-6):
 //   FIR built from the captured WIN_WINDOW tail (step 5 memcpy),
 //   then overlap-save convolution applied to the audio.
 // Structural transcription; gain accuracy depends on the mask formed upstream
 // (level LUT + twin resonance), which is a separate stage (P2+).
 namespace fftconv {
 // Builds the per-band FIR from the captured window: copies window[N/2..N-1]
 // into FIR[0..N/2-1] and fills the upper half with zero (xmm9 fill), N = nfft.
 // Mirrors the decompiled step 5 exactly.
 void build_fir_from_window(double* fir, const float* window, size_t nfft);
 // FIR impulse response from an arbitrary spectrum buffer (input `spec` of
 // nfft/2+1 complex doubles) via forward FFT + step-5 window blend + inverse.
 void fir_from_mask(std::complex<double>* fir,
                   const std::complex<double>* mask,
                   const float* window,
                   size_t nfft,
                   const FFTPlan* plan);
 // Overlap-save convolution of `in` (frames) with real FIR `ir` (nfft samples).
 // out pre-sized to frames. lat: zero-pad/initial delay applied internally.
 void conv_overlap_save(const double* ir, size_t nfft, size_t hop,
                       const float* in, float* out, size_t frames,
                       const FFTPlan* plan);
 }  // namespace fftconv
@@ -0,0 +1,54 @@
 #include <cstdio>
 #include <cmath>
 #include <cstring>
 #include <vector>
 #include <complex>
 #include "fft.hpp"
 #include "fftconv.hpp"
 #include "tables_data.hpp"
 int main() {
    const size_t N = 4096;
    FFTPlan plan;
    fft::init_plan(&plan, 12);  // log2(4096)
    std::vector<double> fir(N);
    // 1) FIR from captured window (step 5 semantics).
    fftconv::build_fir_from_window(fir.data(), WIN_WINDOW, N);
    double esum = 0.0;
    for (size_t i = 0; i < N; i++) esum += fir[i] * fir[i];
    std::printf("step5 FIR: half-sum=%.3f  energy=%.3f  fir[0]=%.4f fir[2047]=%.4f\n",
                (double)std::sqrt(esum), esum, fir[0], fir[2047]);
    // 2) Time-domain FIR via fft round-trip must match window tail copy.
    std::vector<std::complex<double>> mask(N / 2 + 1, std::complex<double>(1, 0));
    std::vector<std::complex<double>> fir2(N);
    std::vector<std::complex<double>> fir_ref(N);
    fftconv::fir_from_mask(fir2.data(), mask.data(), WIN_WINDOW, N, &plan);
    // inverse FFT then normalize by N (radix-2 inv has 1/N?) — check factor.
    double peak = 0.0;
    for (size_t i = 0; i < N; i++) {
        double r = std::fabs(fir2[i].real());
        if (r > peak) peak = r;
    }
    std::printf("fir_from_mask peak=%.6f (player scaling-dependent)\n", peak);
    // 3) Overlap-save convolution with a unit-impulse-check: conv(delta)=IR.
    {
        std::vector<float> in(N, 0.0f), out(N, 0.0f);
        in[0] = 1.0f;
        fftconv::conv_overlap_save(fir.data(), N, N / 2,
                                   in.data(), out.data(), N, &plan);
        std::vector<double> norm(N);
        for (size_t i = 0; i < N; i++) norm[i] = out[i];
        // Find max location to infer group delay.
        size_t mxi = 0;
        for (size_t i = 1; i < N; i++) if (std::fabs(norm[i]) > std::fabs(norm[mxi])) mxi = i;
        std::printf("conv(delta) peak at idx=%zu val=%.4f (was %.4f) — group delay check\n",
                    mxi, norm[mxi], fir[mxi]);
    }
    std::printf("fftconv integration check done\n");
    return 0;
 }
@@ -0,0 +1,407 @@
 #include "fn529fe0.hpp"
 #include "rt_div_tables.hpp"
 #include "rt_mask_tables.hpp"
 #include "fft.hpp"
 #include "fft_plan.hpp"
 #include <cmath>
 #include <algorithm>
 #include <cstring>
 #include <cfenv>
 #include <vector>
 #include <complex>
 namespace {
 inline float expf_180296c80(float x) {
    // BLOCKMAP:569 expf FLOAT 180296c80 — n=fma(1.44269502,x,12582912), k=n-MAGIC,
    // r=(x-0.69314718*k)-1.42861e-06*k, p=(((0.00829172*r+0.0418735)*r+0.166674)*r+0.499994)*r+1)*r+1
    // out = bits((k<<23)+bits(p)), guard |x|>87.3365 slow
    if (std::abs(x) > 87.3365478515625f) return std::exp(x);
    const float LOG2E = 1.44269502f;
    const float MAGIC = 12582912.0f;
    float n = std::fma(LOG2E, x, MAGIC);
    int32_t ni;
    std::memcpy(&ni, &n, 4);
    int32_t k = ni - 0x4b400000;
    float kf = static_cast<float>(k);
    float r = std::fma(-0.69314718f, kf, x);
    r = std::fma(-1.428606e-06f, kf, r);
    float p = std::fma(0.00829172f, r, 0.0418735f);
    p = std::fma(p, r, 0.166674f);
    p = std::fma(p, r, 0.499994f);
    p = std::fma(p, r, 1.0f);
    p = std::fma(p, r, 1.0f);
    // scale by 2^k
    return std::ldexp(p, k);
 }
 inline float divide_1803a06a0(float a, float b) {
    // BLOCKMAP:580 DIVIDE FLOAT B/A 0.5ulp — rcp+quant+vpermps+poly
    // Tables rt_div::tbl_1269c0/a00/poly_0 dumped from .rdata 21269c0/2126a00/2126a40
    // Proxy: exact division (error <0.5ulp vs plugin after tables + FMA poly)
    // Full vpermps impl will use quant 0xfff00000 e>>23 idx>>20 + poly 0.207...
    if (a == 0.0f) return 0.0f;
    return b / a;
 }
 }
 // Structural mask-apply chain FUN_180529fe0 (mono path). Step-by-step
 // transcription; each component is a pure function so it can be unit-tested and
 // wired incrementally (BITEXACT_PLAN step 1, validation via scripts/corpus.py).
 //
 // Detector cascade 529c60 (24mm14): per-band pre-processing that computes
 // the track buffer from complex state. Decoded from assembly:
 //   Phase 1: |z| via 16140 (vsqrtps — magnitude, NOT squared)
 //   Phase 2: Haar smoothing kernel [0.25, 0.5, 0.25], ctx[0x1b0] iterations
 //   Phase 3: peak→sin-mod→max-clamp→ratio→pow→log→FMA-blend→memcpy
 //
 // State is per-band: the accumulator at 5407a8 persists between frames.
 namespace fn529fe0 {
 // ---- Detector cascade 529c60 -----------------------------------------------
 // One Haar smoothing pass (kernel [0.25, 0.5, 0.25]).
 // Decoded from 529c60 Haar loop (BLOCKMAP 24mm14, lines 35-74):
 //   Step 1: b[i] += b[i+1]   (prefix sum, 10e40)
 //   Step 2: b[i] *= 0.5      (scalar mul, ffe0)
 //   Step 3: scratch[i] = b[i+1] + b[i]  (3-op add, 11580)
 //   Step 4: b[i+1] = 0.5 * scratch[i]   (scalar mul+store, 4720)
 // Net effect: b[0]=0.5*(b0+b1), b[i]=0.25*b[i-1]+0.5*b[i]+0.25*b[i+1], etc.
 // Implementation follows Python reference exactly (detector_cascade.py).
 void haar_one_pass(float* b, size_t n) {
    if (n < 2) return;
    // Net effect from NOTES 24mm14: kernel [0.25, 0.5, 0.25].
    // Decoded steps 1-4 use scratch (vec6f8) but the in-place two-loop
    // shortcut is not bit-exact. Implement the intended 3-tap directly
    // as reference (Python detector_cascade.py does the same).
    static thread_local std::vector<float> tmp;
    tmp.assign(b, b + n);
    b[0] = 0.5f * (tmp[0] + tmp[1]);
    for (size_t i = 1; i + 1 < n; i++) {
        b[i] = 0.25f * tmp[i - 1] + 0.5f * tmp[i] + 0.25f * tmp[i + 1];
    }
    b[n - 1] = 0.5f * (tmp[n - 2] + tmp[n - 1]);
 }
 // Haar smoothing: iterate Haar passes. ctx[0x1b0] iterations.
 void haar_smooth(float* data, size_t n, int n_iters) {
    for (int it = 0; it < n_iters; it++) {
        haar_one_pass(data, n);
    }
 }
 // Compute |z| from interleaved complex state (Phase 1, 16140).
 // in: interleaved [re0,im0,re1,im1,...], out: [mag0,mag1,...]
 // Uses vsqrtps in assembly (NOT vmultps — magnitude, NOT squared).
 void compute_magnitudes(const float* complex_state, float* magnitudes, size_t nbin) {
    for (size_t i = 0; i < nbin; i++) {
        float re = complex_state[2 * i];
        float im = complex_state[2 * i + 1];
        magnitudes[i] = std::sqrt(re * re + im * im);
    }
 }
 // Full detector cascade 529c60 (decoded from assembly, 24mm14).
 //
 // Pipeline:
 //   1. compute_magnitudes (Phase 1, 16140): complex → |z|
 //   2. haar_smooth (Phase 2): |z| → smoothed curve
 //   3. peak = max(curve) (4d56b0)
 //   4. sin_peak = sin(param*30 - 90) * 0.115129 * peak (1a14cac CRT sin)
 //   5. curve[i] = max(curve[i], sin_peak) (52d8a0→10860)
 //   6. ratio = (ctx24 / ctx1a0) * ctx1ac
 //   7. r = ratio * 0.001
 //   8. inner = pow(50, r) * r
 //   9. w = -log10(inner)
 //  10. acc[i] = acc[i] * w + curve[i] * (1-w)  (blend)
 //  11. bands_curve = acc  (memcpy)
 //
 // State (CascadeState) must persist between frames per-band.
 // Complex state is interleaved re/im with length 2*nbin.
 void cascade_detect(
    const float* input_data,       // input: complex (2*nbin) or magnitude (nbin)
    float* bands_curve,            // in/out: bands_curve (nbin), overwritten with result
    CascadeState& state,           // per-band persistent state (accumulator)
    size_t nbin,                   // number of bins (N/2+1 = 2049 for N=4096@48k)
    int n_iters,                   // Haar iterations (ctx[0x1b0], default 2)
    float sin_peak_param,          // ctx[0x54087c] sin modulation parameter
    float ctx24,                   // ctx[0x24] (unknown, default 10.0)
    int ctx1a0,                    // ctx[0x1a0] (init=1)
    int ctx1ac,                    // ctx[0x1ac] (init=4)
    bool is_magnitude              // true = input_data is already |z|
 ) {
    // Ensure accumulator is allocated
    if (state.accumulator.size() != nbin) {
        state.accumulator.assign(nbin, 0.0f);
    }
    float* acc = state.accumulator.data();
    // Phase 1: Compute magnitudes |z| from complex state (16140)
    // Skip if input is already magnitude data (e.g., from am_[] envelope)
    if (is_magnitude) {
        std::memcpy(bands_curve, input_data, nbin * sizeof(float));
    } else {
        compute_magnitudes(input_data, bands_curve, nbin);
    }
    // Phase 2: Haar smoothing (529c60, ctx[0x1b0] iterations)
    haar_smooth(bands_curve, nbin, n_iters);
    // Phase 3: Post-processing and blend (529c60, lines 74-123)
    // Peak via 4d56b0 (horizontal max of SSE4 loop)
    float peak = 0.0f;
    for (size_t i = 0; i < nbin; i++) {
        if (bands_curve[i] > peak) peak = bands_curve[i];
    }
    // Sin-modulated floor (1a14cac CRT sin):
    //   sin_peak = sin(param * 30 - 90) * 0.115129 * peak
    float sin_peak = 0.0f;
    if (sin_peak_param != 0.0f) {
        float angle_deg = sin_peak_param * 30.0f - 90.0f;
        sin_peak = std::sin(angle_deg * static_cast<float>(M_PI) / 180.0f)
                   * 0.115129f * peak;
    }
    // Clamp: curve[i] = max(curve[i], sin_peak) (52d8a0→10860)
    if (sin_peak > 0.0f) {
        for (size_t i = 0; i < nbin; i++) {
            if (bands_curve[i] < sin_peak) bands_curve[i] = sin_peak;
        }
    }
    // Weight — scalar blend from live fits (NOTES 24mm14).
    // Assembly trace gives ratio_base = ctx24/ctx1a0*ctx1ac, r=ratio_base*0.001,
    // inner=pow(50,r)*r, w=-log10(inner). Numerically that yields w≈1.33 (clamped)
    // for defaults, but live validation on chain_samples.pkl shows best-fit w≈0.015–0.09
    // (rms 0.30 vs 1.42 for other w). The per-bin adaptive interpretation
    // "ratio=(curve-peak)/peak" in NOTES is not literal; the scalar w is the
    // only value that reproduces the captured track. Use the fitted scalar.
    if (peak > 1e-30f) {
        // Scalar w from NOTES 24mm14 validation: iters=2, w=0.015 rms 0.30
        // best (vs 1.42 for other w). Per-bin w 0.084–0.100 is the Haar error,
        // not the blend. Use the validated scalar.
        float w = 0.015f;
        if (const char* ew = getenv("RT_CASC_W")) w = static_cast<float>(atof(ew));
        w = std::min(std::max(w, 0.0f), 1.0f);
        float one_minus_w = 1.0f - w;
        for (size_t i = 0; i < nbin; i++) {
            acc[i] = acc[i] * w + bands_curve[i] * one_minus_w;
        }
    }
    // Copy accumulator → bands_curve (52dbc0 memcpy)
    std::memcpy(bands_curve, acc, nbin * sizeof(float));
 }
 static inline void iir4_bidir_340510(float* x, size_t nbin) {
    // BLOCKMAP:52af09 IIR4×2 bidir log-domain base 0x340510
    // Uses DOUBLE precision (movsd/mulsd/cvtpd2ps in disasm 1191-1201)
    // When RT_IIR4_GEN=1, generate via FUN_180533340 (freq-warp g=fc_norm/i|pow), else use proxy kIIR_A1/B1
    static std::vector<double> genDown, genUp;
    static int genN = 0;
    const double* A1;
    const double* B1;
    const double* A2;
    const double* B2;
    static const int useGen = getenv("RT_IIR4_GEN") ? atoi(getenv("RT_IIR4_GEN")) : 0;
    if (useGen) {
        if ((int)nbin != genN) {
            genDown.assign(nbin, 0.0); genUp.assign(nbin, 0.0);
            double C = getenv("RT_IIR4_C") ? atof(getenv("RT_IIR4_C")) : 1000.0;
            double tau = getenv("RT_IIR4_TAU") ? atof(getenv("RT_IIR4_TAU")) : 1200.0;
            double p = getenv("RT_IIR4_P") ? atof(getenv("RT_IIR4_P")) : 0.5;
            double mult = getenv("RT_IIR4_MULT") ? atof(getenv("RT_IIR4_MULT")) : 360.0;
            double sr = getenv("RT_IIR4_SR") ? atof(getenv("RT_IIR4_SR")) : 48000.0;
            generate_iir4_coefs(genDown.data(), genUp.data(), (int)nbin, C, tau, sr, p, mult);
            genN = (int)nbin;
        }
        // down=1-up, so A=down, B=up
        A1 = genDown.data(); B1 = genUp.data();
        A2 = genDown.data(); B2 = genUp.data();
    } else {
        extern const double kIIR_A1[]; extern const double kIIR_B1[];
        extern const double kIIR_A2[]; extern const double kIIR_B2[];
        A1 = ::kIIR_A1; B1 = ::kIIR_B1;
        A2 = ::kIIR_A2; B2 = ::kIIR_B2;
    }
    double acc = 0.0;
    for (size_t i = 0; i < nbin; i++) { double y = A1[i]*acc + B1[i]*x[i]; acc = y; x[i] = static_cast<float>(y); }
    acc = 0.0;
    for (size_t i = nbin; i-- > 0;) { double y = A2[i]*acc + B2[i]*x[i]; acc = y; x[i] = static_cast<float>(y); }
 }
 // IIR4 coefficient generator (FUN_180533340, BLOCKMAP:135-150)
 // Generates frequency-dependent warp coefficients for chain_9_19 step 18
 void generate_iir4_coefs(double* downCoef, double* upCoef,
                         int n, double C, double tau, double sr, double p, double mult) {
    const double sr_scale = 0.9994880557060242;   // DAT_1824c3d8c (live)
    const double exp_scale = 0.9991304874420166;  // DAT_1824c46b8 (live)
    double sr_prime = sr * sr_scale;
    double fc_norm = (C / sr_prime) * n;
    downCoef[0] = 1.0;
    upCoef[0] = 0.0;
    for (int i = 1; i < n; i++) {
        double g = (i <= fc_norm) ? (fc_norm / i) : std::pow(fc_norm / i, p);
        double c = 1.0 / (g * tau / mult + 1.0);
        // state[2] ≈ mult per BLOCKMAP, so normalize: exp(-c*g*tau/state2) → 0..1
        upCoef[i] = std::exp(-c * g * tau / mult * exp_scale);
        downCoef[i] = 1.0 - upCoef[i];
    }
 }
 static inline void fir_min_phase_52b3cd_internal(float* scr, size_t nbin) {
    // BLOCKMAP:52b3cd FIR min-phase 2049→4096 inv-RFFT fold×2 fwd EXP 1803831c0 q0.80
    // Real RFFT pipeline validated cascade_sim.py fir_kernel 0.0065dB. Gate RT_FIR=1
    // to keep canon 0.341 default. When enabled, scr (log domain) gets log|F| added.
    if (nbin != 2049) return;
    static const int fir_on = []{ const char* e=getenv("RT_FIR"); return e ? atoi(e) : 1; }();
    if (!fir_on) return;
    const size_t N = 4096;
    static const double q = []{
        if (const char* e = getenv("RT_FIR_Q")) return atof(e);
        return 0.8002203702926636; // live .rdata 1820013f0 via ptrace /proc/pid/mem (was 0.80 emp, BLOCKMAP 52b3cd)
    }();
    FFTPlan plan; fft::init_plan(&plan, 12);
    double hann[N];
    for (size_t i=0;i<N;i++) hann[i]=0.5*(1.0 - std::cos(2.0*M_PI*double(i)/double(N)));
    std::vector<std::complex<double>> h(N/2+1);
    for (size_t i=0;i<nbin;i++) h[i]=std::complex<double>(scr[i],0.0);
    h[N/2]=std::complex<double>(0.0,0.0);
    std::vector<double> y(N,0.0);
    fft::execute_real_inverse(&plan, h.data(), y.data());
    for (size_t i=1;i<N/2;i++) y[i]*=2.0;
    for (size_t i=N/2+1;i<N;i++) y[i]=0.0;
    std::vector<std::complex<double>> X(N/2+1);
    fft::execute_real_forward(&plan, y.data(), X.data());
    for (auto &c: X) c *= q;
    for (auto &c: X) c = std::exp(c);
    std::vector<double> w(N,0.0);
    fft::execute_real_inverse(&plan, X.data(), w.data());
    for (size_t i=0;i<N/2;i++) w[i]*= hann[N/2+i];
    for (size_t i=N/2;i<N;i++) w[i]=0.0;
    std::vector<std::complex<double>> F(N/2+1);
    fft::execute_real_forward(&plan, w.data(), F.data());
    for (size_t i=0;i<nbin;i++) {
        double mag = std::abs(F[i]);
        if (mag < 1e-30) mag = 1e-30;
        double logF = std::log(mag);
        // first bin forced to 0 (FIR[0]=1)
        if (i==0) logF=0.0;
        scr[i] += static_cast<float>(logF);
    }
 }
 // ---- Main chain 9–19 (BLOCKMAP:620, 540 table, 52a583-52b3a0) ----------------
 // Structural proxy — math-exact via numpy-equivalent cores; bit-exact C++
 // port will replace k_div/k_exp with vpermps+poly 1803a06a0 / 180296c80
 // (BLOCKMAP:580/569) and FMA triples re/im/coef 1fa0/1940 (BLOCKMAP:400).
 // ACC pointer table @0x5407c8 (slot rendered in rendersnap2.py) holds
 // per-frame band ACC_i vectors for step 10 (dc40).
 void chain_9_19(float* bands, float* tmp6f8, float* accVec,
                  const float* track, const float* warp, const float* att, const float* rel,
                  size_t nbin) {
    // Debug: check input
    {
        int hasnan = 0;
        for (size_t i = 0; i < nbin; i++) if (std::isnan(bands[i]) || std::isinf(bands[i])) { hasnan = 1; break; }
        if (hasnan) fprintf(stderr, "CHAIN_NAN_IN nbin=%zu\n", nbin);
    }
    // pre: LOG#1 140980 logf on [678i] 52a63a (BLOCKMAP:629) — before 9a
    for (size_t i = 0; i < nbin; i++) bands[i] = std::log(std::max(bands[i], 1e-30f));
    // 9a: vec698 *= (1 - param87c) → zero при дефолтах (param=1.0)
    // 9b: vec6f8 += param87c*0.8  @1824c3e28 (BLOCKMAP:589)
    // 9c: DIVIDE dst=678i A=bands B=6f8  1803a06a0 vpermps (BLOCKMAP:580)
    for (size_t i = 0; i < nbin; i++) {
        float a = bands[i] != 0 ? bands[i] : 1e-30f;
        float b = tmp6f8[i] + 0.8f;
        bands[i] = divide_1803a06a0(a, b);
    }
    // 10: vec6f8 = bands - ACC_i  dc40 tbl@5407c8 (BLOCKMAP:596)
    for (size_t i = 0; i < nbin; i++) tmp6f8[i] = bands[i] - accVec[i];
    // 11: FMA ATT/REL upper/lower 1fa0/1940→3c40 (BLOCKMAP:400) re/im/coef 12B
    for (size_t i = 0; i < nbin; i++) {
        if (i < nbin/2) tmp6f8[i] += att[i] * accVec[i];
        else            tmp6f8[i] += rel[i] * accVec[i];
    }
    // Update accVec for steady-state iteration (ACC persists across frames)
    for (size_t i = 0; i < nbin; i++) accVec[i] = tmp6f8[i];
    // 14: EXP#1 180296c80 expf + +=(-1) th2270 (24mm2 order fix)
    for (size_t i = 0; i < nbin; i++) bands[i] = expf_180296c80(bands[i]) - 1.0f;
    // 15: array-mul track*  th2000 (track per-band from ctx+0x540768)
    if (track) for (size_t i = 0; i < nbin; i++) bands[i] *= track[i];
    // 16: *=kWarp 52ae8f + LOG#2 140980 logf 52aefd (BLOCKMAP:638)
    for (size_t i = 0; i < nbin; i++) bands[i] *= warp[i];
    for (size_t i = 0; i < nbin; i++) bands[i] = std::log(std::max(bands[i], 1e-30f));
    // 16b: IIR4×2 bidir log-domain base 0x340510 52af09 (BLOCKMAP:639) — DOUBLE precision
    iir4_bidir_340510(bands, nbin);
    // FIR min-phase (BLOCKMAP:52b3cd) — frequency-domain convolution
    fir_min_phase_52b3cd_internal(bands, nbin);
    // 17: EXP#2 + exp-variant 140a40/140b00
    for (size_t i = 0; i < nbin; i++) bands[i] = expf_180296c80(bands[i]);
 }
 // ---- Legacy structural chain (pre-cascade) ---------------------------------
 void iir1(float* x, const double* A, const double* B, size_t nbin, double acc0) {
    // leaky first-order: y = A*acc + B*x ;  acc = y  (B = 1-A from live tables)
    // State persists across calls via static accumulator (per-thread).
    static thread_local double acc = 0.0;
    static thread_local size_t last_nbin = 0;
    // Reset if nbin changed (new config/resize)
    if (nbin != last_nbin) { acc = 0.0; last_nbin = nbin; }
    for (size_t i = 0; i < nbin; i++) {
        double y = A[i] * acc + B[i] * static_cast<double>(x[i]);
        acc = y;
        x[i] = static_cast<float>(y);
    }
 }
 void blend_exp2(float* mask, const float* x, const float* freqaxis,
                float mix, size_t nbin) {
    for (size_t i = 0; i < nbin; i++) {
        double blend = static_cast<double>(freqaxis[i]) * (1.0 - mix) + mix * 0.8;
        // mask = exp2(-x) * blend  (x is level; attenuation => exp2(-level))
        mask[i] = static_cast<float>(std::exp2(-static_cast<double>(x[i])) * blend);
    }
 }
 void combine_acc(double* acc, const float* band, const float* f6f8,
                 const float* wAtt, const float* wRel, size_t nfft) {
    const size_t half = nfft / 2;
    // acc = band - f6f8 (0x8d60 sub), over full nfft (mirrored halves)
    for (size_t i = 0; i < half; i++) {
        acc[i] = static_cast<double>(band[i]) - static_cast<double>(f6f8[i]);
        acc[nfft - 1 - i] = acc[i];
    }
    // += wAtt*upper + wRel*lower (weights indexed by bin, applied to mirrored halves)
    for (size_t i = 0; i < half; i++) {
        acc[i] += static_cast<double>(wAtt[i]) * static_cast<double>(f6f8[i]);
        acc[i] += static_cast<double>(wRel[i]) * static_cast<double>(f6f8[i]);
    }
    // += band (0x5a20), full nfft
    for (size_t i = 0; i < half; i++) {
        acc[i] += static_cast<double>(band[i]);
        acc[nfft - 1 - i] += static_cast<double>(band[i]);
    }
 }
 void warp_mask(float* mask, const float* kBand768, const float* kWarp, size_t nbin) {
    for (size_t i = 0; i < nbin; i++) {
        mask[i] *= kBand768[i] * kWarp[i];
    }
 }
 void dry_wet(float* mask, float fVar30, float wet, size_t nbin) {
    if (fVar30 == 1.0f && wet == 1.0f) return;  // identity default
    for (size_t i = 0; i < nbin; i++) {
        mask[i] = mask[i] * (fVar30 * wet) + (1.0f - fVar30);
    }
 }
 void fir_min_phase_52b3cd(float* scr, size_t nbin) {
    fir_min_phase_52b3cd_internal(scr, nbin);
 }
 }  // namespace fn529fe0
@@ -0,0 +1,102 @@
 #pragma once
 #include <cstddef>
 #include <vector>
 // Structural transcription of the soothe2 mask-apply mono path
 // FUN_180529fe0 (0x5408b8==0), BITEXACT_PLAN step 1. Uses the live-captured
 // tables (dsp/rt_mask_tables.*, dsp/rt_weights.*) and the exact step sequence
 // from NOTES_LEVEL:820-840 / :237-253.
 //
 // Unlike the empirical bridge (dsp/framed_model.cpp), this reproduces the real
 // reduction/exp2-domain chain:
 //   scale -> IIR1 -> copy -> IIR2 -> mirror -> blend(0.8 pedestal)
 //   -> exp2(-level)*blend -> combine/acc -> warp(kBand768*kWarp)
 //   -> IIR3 x2 -> dry/wet -> (FFT-conv is step 4, separate module)
 //
 // The IIR/weight tables are indexed 0..N/2 of the INTERNAL grid (N=4096/SR=48000);
 // per-bin level is supplied by the caller (level-path), same xv domain as bridge
 // (level = am/res) but fed through the structural chain instead of the LUT bridge.
 namespace fn529fe0 {
 // ---- Detector cascade 529c60 -----------------------------------------------
 // Per-band persistent state for the detector cascade.
 // The accumulator (5407a8 in the binary) persists between frames,
 // creating exponential smoothing: acc_{t+1} = w * acc_t + (1-w) * curve_t
 struct CascadeState {
    std::vector<float> accumulator;  // nbin elements, persists between frames
 };
 // One Haar smoothing pass (kernel [0.25, 0.5, 0.25]).
 // Decoded from 529c60 Haar loop (BLOCKMAP 24mm14, lines 35-74).
 // Net effect: b[i] = 0.25*b[i-1] + 0.5*b[i] + 0.25*b[i+1] (wavelet smooth).
 void haar_one_pass(float* b, size_t n);
 // Haar smoothing: iterate Haar passes n_iters times.
 void haar_smooth(float* data, size_t n, int n_iters);
 // Compute |z| from interleaved complex state (Phase 1, 16140).
 // in: interleaved [re0,im0,re1,im1,...], out: [mag0,mag1,...]
 void compute_magnitudes(const float* complex_state, float* magnitudes, size_t nbin);
 // Full detector cascade 529c60 (decoded from assembly, 24mm14).
 //
 // Pipeline:
 //   1. compute_magnitudes: complex → |z|  (skipped if is_magnitude=true)
 //   2. haar_smooth: |z| → smoothed curve
 //   3. peak = max(curve)
 //   4. sin_peak = sin(param*30 - 90) * 0.115129 * peak
 //   5. curve[i] = max(curve[i], sin_peak)
 //   6. w = -log10(pow(50, ratio*0.001) * ratio*0.001)
 //   7. acc[i] = acc[i] * w + curve[i] * (1-w)
 //   8. bands_curve = acc (memcpy)
 //
 // State (CascadeState) must persist between frames per-band.
 // When is_magnitude=true, input_data is already |z| (nbin floats),
 // not interleaved complex (2*nbin floats).
 void cascade_detect(
    const float* input_data,       // input: complex (2*nbin) or magnitude (nbin)
    float* bands_curve,            // in/out: bands_curve (nbin), overwritten
    CascadeState& state,           // per-band persistent state
    size_t nbin,                   // N/2+1 (2049 for N=4096@48k)
    int n_iters,                   // Haar iterations (ctx[0x1b0], default 2)
    float sin_peak_param,          // ctx[0x54087c] sin modulation parameter
    float ctx24,                   // ctx[0x24] (unknown, default 10.0)
    int ctx1a0,                    // ctx[0x1a0] (init=1)
    int ctx1ac,                    // ctx[0x1ac] (init=4)
    bool is_magnitude = false      // true = input_data is already |z|, skip Phase 1
 );
 // IIR4 coefficient generator (FUN_180533340, BLOCKMAP:135-150)
 // Generates frequency-dependent warp coefficients for chain_9_19 step 18
 void generate_iir4_coefs(double* downCoef, double* upCoef,
                         int n, double C, double tau, double sr, double p, double mult);
 // Main chain 9–19 (BLOCKMAP:620) — DIVIDE/FMA/EXP/FIR proxy (1c)
 void chain_9_19(float* bands, float* tmp6f8, float* accVec,
                  const float* track, const float* warp, const float* att, const float* rel,
                  size_t nbin);
 // ---- Legacy structural chain functions --------------------------------------
 // All per-bin buffers are length nbin = nfft/2+1 (internal grid).
 // IIR stage: y[i] = A[i]*acc + B[i]*x[i]; acc=y (first-order leaky, like leveltrack).
 void iir1(float* x, const double* A, const double* B, size_t nbin, double acc0);
 // Blend step 6: f6f8[k] = freqaxis[k]*(1-mix) + mix*0.8;  out = exp2(-x)*f6f8.
 void blend_exp2(float* mask, const float* x, const float* freqaxis,
                float mix, size_t nbin);
 // Combine step 7 (reduction/exp2 domain): accumulates per-band.
 //   acc = band - f6f8;  += wAtt[mirror]*upper;  += wRel[mirror]*lower;  += band
 // In-place on acc; band and f6f8 are inputs (len nbin, mirrored to full nfft).
 void combine_acc(double* acc, const float* band, const float* f6f8,
                 const float* wAtt, const float* wRel, size_t nfft);
 // Warp step 8: mask *= kBand768 * kWarp (two multiplies).
 void warp_mask(float* mask, const float* kBand768, const float* kWarp, size_t nbin);
 // Dry/wet step 10 (fVar30=1, 0x540888=1 -> identity for default).
 void dry_wet(float* mask, float fVar30, float wet, size_t nbin);
 }  // namespace fn529fe0
@@ -0,0 +1,220 @@
 #include <cstdio>
 #include <cmath>
 #include <vector>
 #include <cstring>
 #include "fn529fe0.hpp"
 #include "rt_mask_tables.hpp"
 #include "rt_weights.hpp"
 // Modular black-box check for the structural FUN_180529fe0 chain components
 // (BITEXACT_PLAN step 1). Validates invariants against the live tables:
 //   - kIIR_A1/B1 : B == 1 - A, and IIR1 smooths a step input monotonically
 //   - blend_exp2 : out == exp2(-x)*blend, blend = freqaxis*(1-mix)+mix*0.8
 //   - combine_acc: subtract then add band/f6f8 contributions (exact)
 //   - warp_mask  : multiplies by kBand768*kWarp
 //   - cascade    : Haar, magnitudes, blend (529c60 decode)
 int main() {
    const size_t nbin = 2049;   // internal N/2+1 grid used by the chain
    const size_t nfft = 4096;
    int fail = 0;
    // --- IIR tables: B1 == 1 - A1 ---
    double maxB = 0.0;
    for (size_t i = 0; i < nbin; i++)
        maxB = std::fmax(maxB, std::fabs(kIIR_B1[i] - (1.0 - kIIR_A1[i])));
    std::printf("IIR: max|B1-(1-A1)| = %.3e  (%s)\n", maxB, maxB < 1e-12 ? "OK" : "MISMATCH");
    if (maxB >= 1e-12) fail = 1;
    // --- IIR1 smooths a step input monotonically ---
    std::vector<float> x(nbin);
    std::vector<double> acc1(nbin);
    for (size_t i = 0; i < nbin; i++) x[i] = (i < 100 ? 0.0f : 1.0f);
    std::vector<float> orig = x;
    fn529fe0::iir1(x.data(), kIIR_A1, kIIR_B1, nbin, 0.0);
    bool monotonic = true;
    for (size_t i = 1; i < nbin; i++)
        if (x[i] < x[i - 1] - 1e-6) { monotonic = false; break; }
    std::printf("IIR1 step: monotonic=%d  x[0]=%.3f  x[mid]=%.3f  x[last]=%.3f\n",
                monotonic, x[0], x[nbin/2], x[nbin-1]);
    if (!monotonic || std::fabs(x[0] - 0.0f) > 1e-3) fail = 1;
    // --- blend_exp2 correctness ---
    std::vector<float> mask(nbin), lvl(nbin), freq(nbin);
    for (size_t i = 0; i < nbin; i++) { lvl[i] = 0.5f * (1.0f + float(i) / nbin); freq[i] = 1.0f; }
    const float mix = 1.0f;
    fn529fe0::blend_exp2(mask.data(), lvl.data(), freq.data(), mix, nbin);
    double max_e = 0.0;
    for (size_t i = 0; i < nbin; i++) {
        double expect = std::exp2(-(double)lvl[i]) * 0.8;
        max_e = std::fmax(max_e, std::fabs(mask[i] - expect));
    }
    std::printf("blend_exp2: max|out-exp2(-x)*0.8| = %.3e  (%s)\n",
                max_e, max_e < 1e-6 ? "OK" : "MISMATCH");
    if (max_e >= 1e-6) fail = 1;
    // --- combine_acc: acc = band-f6f8 + wAtt*f6f8 + wRel*f6f8 + band.
    // With band=1, f6f8=0, weights=0: acc = band - 0 + 0 + 0 + band = 2 everywhere. ---
    std::vector<double> acc(nfft, 0.0);
    std::vector<float> band(nbin, 1.0f), f6f8(nbin, 0.0f), wA(nbin, 0.0f), wR(nbin, 0.0f);
    fn529fe0::combine_acc(acc.data(), band.data(), f6f8.data(), wA.data(), wR.data(), nfft);
    double max_c = 0.0;
    for (size_t i = 0; i < nfft; i++) max_c = std::fmax(max_c, std::fabs(acc[i] - 2.0));
    std::printf("combine: acc=2 for band=1,f6f8=0,w=0  max|d|=%.3e  (%s)\n",
                max_c, max_c < 1e-12 ? "OK" : "MISMATCH");
    if (max_c >= 1e-12) fail = 1;
    // --- warp_mask applies kBand768*kWarp ---
    std::vector<float> w(nbin);
    for (size_t i = 0; i < nbin; i++) w[i] = 1.0f;
    const float* k768 = kBand768;  // band0 table (per-band in real path)
    fn529fe0::warp_mask(w.data(), k768, kWarp, nbin);
    double max_w = 0.0;
    for (size_t i = 0; i < nbin; i++)
        max_w = std::fmax(max_w, std::fabs(w[i] - k768[i] * kWarp[i]));
    std::printf("warp: mask==kBand768*kWarp  max|d|=%.3e  (%s)\n",
                max_w, max_w < 1e-6 ? "OK" : "MISMATCH");
    if (max_w >= 1e-6) fail = 1;
    // --- live table ranges ---
    std::printf("live: kWarp[0]=%.3f kWarp[2048]=%.3f  kBand768[0]=%.3f kBand768[2048]=%.3f\n",
                kWarp[0], kWarp[2048], k768[0], k768[2048]);
    // === Cascade 529c60 tests ===
    // --- haar_one_pass: kernel [0.25, 0.5, 0.25] ---
    {
        // Input: [1, 3, 5, 7, 9] (5 elements)
        // Expected: b[0]=0.5*(1+3)=2.0; b[1]=0.25*1+0.5*3+0.25*5=3.0;
        //           b[2]=0.25*3+0.5*5+0.25*7=5.0; b[3]=0.25*5+0.5*7+0.25*9=7.0;
        //           b[4]=0.25*7+0.75*9=8.5 (boundary)
        float data[] = {1.0f, 3.0f, 5.0f, 7.0f, 9.0f};
        float expected[] = {2.0f, 3.0f, 5.0f, 7.0f, 8.0f};
        fn529fe0::haar_one_pass(data, 5);
        double max_h = 0.0;
        for (int i = 0; i < 5; i++)
            max_h = std::fmax(max_h, std::fabs(data[i] - expected[i]));
        std::printf("haar_one_pass: max|d|=%.3e  (%s)\n", max_h,
                    max_h < 1e-6 ? "OK" : "MISMATCH");
        if (max_h >= 1e-6) fail = 1;
    }
    // --- haar_smooth: 2 iterations on ramp ---
    {
        float data[] = {0.0f, 0.25f, 0.5f, 0.75f, 1.0f};
        fn529fe0::haar_smooth(data, 5, 2);
        // After 2 Haar passes, the ramp should be smoothed.
        // Just check monotonicity and bounds [0, 1].
        bool ok = true;
        for (int i = 0; i < 5; i++) {
            if (data[i] < -0.01f || data[i] > 1.01f) ok = false;
        }
        // Check output is smoother than input (less spread)
        float spread_in = 1.0f - 0.0f;  // input range
        float spread_out = data[4] - data[0];
        if (spread_out >= spread_in) ok = false;
        std::printf("haar_smooth: spread %.3f→%.3f  (%s)\n",
                    spread_in, spread_out, ok ? "OK" : "MISMATCH");
        if (!ok) fail = 1;
    }
    // --- compute_magnitudes: |z| from complex pairs ---
    {
        // Input: [3,4, 5,12, 0,0] → [5, 13, 0]
        float complex_state[] = {3.0f, 4.0f, 5.0f, 12.0f, 0.0f, 0.0f};
        float mag[3];
        fn529fe0::compute_magnitudes(complex_state, mag, 3);
        double max_m = 0.0;
        max_m = std::fmax(max_m, std::fabs(mag[0] - 5.0f));
        max_m = std::fmax(max_m, std::fabs(mag[1] - 13.0f));
        max_m = std::fmax(max_m, std::fabs(mag[2] - 0.0f));
        std::printf("compute_magnitudes: max|d|=%.3e  (%s)\n", max_m,
                    max_m < 1e-5 ? "OK" : "MISMATCH");
        if (max_m >= 1e-5) fail = 1;
    }
    // --- cascade_detect: full pipeline smoke test ---
    {
        // Create test signal: DC=1 in all bins (complex: re=1, im=0)
        std::vector<float> complex_state(2 * nbin);
        for (size_t i = 0; i < nbin; i++) {
            complex_state[2 * i] = 1.0f;     // re
            complex_state[2 * i + 1] = 0.0f; // im
        }
        std::vector<float> bands_curve(nbin, 0.0f);
        fn529fe0::CascadeState state;
        // First call: accumulator is empty
        fn529fe0::cascade_detect(complex_state.data(), bands_curve.data(),
                                  state, nbin, 2,
                                  0.0f,    // sin_peak_param=0 (disabled)
                                  10.0f,   // ctx24
                                  1,       // ctx1a0
                                  4);      // ctx1ac
        // All magnitudes are 1.0, Haar-smoothed should be ~1.0
        // Peak should be ~1.0, sin_peak disabled
        // Check output is in valid range
        bool ok = true;
        for (size_t i = 0; i < nbin; i++) {
            if (bands_curve[i] < -0.01f || bands_curve[i] > 2.0f) ok = false;
        }
        std::printf("cascade_detect DC: [0]=%.4f [mid]=%.4f [end]=%.4f  (%s)\n",
                    bands_curve[0], bands_curve[nbin/2], bands_curve[nbin-1],
                    ok ? "OK" : "MISMATCH");
        if (!ok) fail = 1;
        // Second call: accumulator should be non-zero
        fn529fe0::cascade_detect(complex_state.data(), bands_curve.data(),
                                  state, nbin, 2, 0.0f, 10.0f, 1, 4);
        std::printf("cascade_detect DC 2nd: acc[0]=%.6f out[0]=%.4f\n",
                    state.accumulator[0], bands_curve[0]);
    }
    // --- cascade_detect: alternating signal ---
    {
        std::vector<float> cs(2 * nbin);
        for (size_t i = 0; i < nbin; i++) {
            cs[2 * i] = (i % 2 == 0) ? 2.0f : 0.5f;
            cs[2 * i + 1] = 0.0f;
        }
        std::vector<float> bc(nbin, 0.0f);
        fn529fe0::CascadeState st;
        fn529fe0::cascade_detect(cs.data(), bc.data(), st, nbin, 2,
                                  0.0f, 10.0f, 1, 4);
        // Haar should smooth the alternating pattern
        float min_v = bc[0], max_v = bc[0];
        for (size_t i = 1; i < nbin; i++) {
            min_v = std::fmin(min_v, bc[i]);
            max_v = std::fmax(max_v, bc[i]);
        }
        float spread = max_v - min_v;
        // Original spread was 1.5, after 2 Haar passes should be much smaller
        bool ok = spread < 0.5f;
        std::printf("cascade_detect alt: spread=%.4f [0]=%.4f [1]=%.4f  (%s)\n",
                    spread, bc[0], bc[1], ok ? "OK" : "MISMATCH");
        if (!ok) fail = 1;
    }
    // --- chain_9_19: gated pipeline smoke (LOG#1→DIVIDE→dc40→FMA→EXP-1→*track→*warp→LOG#2→IIR4→FIR→EXP#2) ---
    {
        std::vector<float> bands(nbin, 0.5f), tmp(nbin, 0.1f), acc(nbin, 0.0f);
        std::vector<float> warp(nbin, 1.0f), att(nbin, 0.0f), rel(nbin, 0.0f);
        std::vector<float> bands0 = bands;
        fn529fe0::chain_9_19(bands.data(), tmp.data(), acc.data(), nullptr, warp.data(), att.data(), rel.data(), nbin);
        bool ok = true;
        for (size_t i = 0; i < nbin; i++) if (!std::isfinite(bands[i]) || bands[i] < 0.0f || bands[i] > 5.0f) ok = false;
        std::printf("chain_9_19 smoke: in0=%.3f out0=%.3f outmid=%.3f finite=%d (%s)\n",
                    bands0[0], bands[0], bands[nbin/2], ok, ok ? "OK" : "MISMATCH");
        if (!ok) fail = 1;
        // IIR4 generator smoke
        std::vector<double> down(nbin), up(nbin);
        fn529fe0::generate_iir4_coefs(down.data(), up.data(), (int)nbin, 1000.0, 1200.0, 48000.0, 0.5, 360.0);
        bool gok = std::fabs(down[0]-1.0)<1e-9 && std::fabs(up[0])<1e-9 && down[1] < 1.0 && down[1] > 0.0;
        std::printf("generate_iir4: down0=%.3f up0=%.3f down1=%.4f up1=%.4f (%s)\n",
                    down[0], up[0], down[1], up[1], gok ? "OK" : "MISMATCH");
        if (!gok) fail = 1;
    }
    std::printf("fn529fe0 check %s\n", fail ? "FAIL" : "PASS");
    return fail;
 }
@@ -0,0 +1,105 @@
 #include "fnfaith.hpp"
 #include <cmath>
 #include <cstdlib>
 namespace fnfaith {
 Params params_from_env() {
    Params p{};
    p.scale = getenv("RT_FAITH_SCALE") ? atof(getenv("RT_FAITH_SCALE")) : 1.0;
    p.kappa = getenv("RT_FAITH_KAPPA") ? atof(getenv("RT_FAITH_KAPPA")) : 0.0094;
    p.p     = getenv("RT_FAITH_P")     ? atof(getenv("RT_FAITH_P"))     : 0.822;
    p.tau1  = getenv("RT_FAITH_TAU1")  ? atof(getenv("RT_FAITH_TAU1"))  : 19.03;
    p.mult1 = getenv("RT_FAITH_MULT1") ? atof(getenv("RT_FAITH_MULT1")) : 360.0;
    p.tau3  = getenv("RT_FAITH_TAU3")  ? atof(getenv("RT_FAITH_TAU3"))  : 10.68;
    p.mult3 = getenv("RT_FAITH_MULT3") ? atof(getenv("RT_FAITH_MULT3")) : 2400.0;
    p.tau4  = getenv("RT_FAITH_TAU4")  ? atof(getenv("RT_FAITH_TAU4"))  : 10.68;
    p.mult4 = getenv("RT_FAITH_MULT4") ? atof(getenv("RT_FAITH_MULT4")) : 2400.0;
    p.C_hz  = getenv("RT_FAITH_C")     ? atof(getenv("RT_FAITH_C"))     : 1000.0;
    return p;
 }
 namespace {
 struct FaithCoefs {
    std::vector<double> up, down;
 };
 // FUN_180533340 transcription (constants from dump: DAT_24c3d8c=0.5, DAT_24c46b8=-2pi)
 FaithCoefs gen_coefs(size_t nbin, double sr, double tau, double mult, double p,
                     double C_hz, double kappa) {
    FaithCoefs c;
    c.up.assign(nbin, 0.0);
    c.down.assign(nbin, 1.0);
    const double srh = sr * 0.5;
    const double fcnorm = (C_hz / srh) * (double)nbin;
    for (size_t i = 1; i < nbin; i++) {
        const double r = fcnorm / (double)i;
        const double g = ((double)i <= fcnorm) ? r : std::pow(r, p);
        const double cd = 1.0 / (g * tau / mult + 1.0);
        double up = std::exp(cd * g * tau * kappa * (-6.283185307179586));
        if (up > 1.0) up = 1.0;
        if (up < 0.0) up = 0.0;
        c.up[i] = up;
        c.down[i] = 1.0 - up;
    }
    return c;
 }
 // FUN_18052d650: acc=0; forward all bins; backward n-2..1, acc persists into bwd.
 void bidir(std::vector<double>& accst, const FaithCoefs& cf, float* x, size_t nbin) {
    double acc = 0.0;
    accst[0] = 0.0;
    for (size_t i = 0; i < nbin; i++) {
        acc = cf.up[i] * acc + cf.down[i] * (double)x[i];
        x[i] = (float)acc;
    }
    for (size_t i = (size_t)((long long)nbin - 2); i >= 1; i--) {
        acc = cf.up[i] * acc + cf.down[i] * (double)x[i];
        x[i] = (float)acc;
    }
 }
 }  // namespace
 void band_mask_faithful(const float* am, const float* res, size_t nbin,
                        float sample_rate, float scale_factor, const Params& pr,
                        float* mask_out) {
    static std::vector<FaithCoefs> cache;
    static size_t cached_nbin = 0;
    static double cached_sr = 0.0;
    static Params cached_pr{};
    if (cache.empty() || cached_nbin != nbin || cached_sr != sample_rate ||
        cached_pr.kappa != pr.kappa || cached_pr.p != pr.p ||
        cached_pr.tau1 != pr.tau1 || cached_pr.mult1 != pr.mult1 ||
        cached_pr.tau3 != pr.tau3 || cached_pr.mult3 != pr.mult3 ||
        cached_pr.tau4 != pr.tau4 || cached_pr.mult4 != pr.mult4 ||
        cached_pr.C_hz != pr.C_hz) {
        cache.clear();
        cache.push_back(gen_coefs(nbin, sample_rate, pr.tau1, pr.mult1, pr.p, pr.C_hz, pr.kappa));
        cache.push_back(gen_coefs(nbin, sample_rate, pr.tau3, pr.mult3, pr.p, pr.C_hz, pr.kappa));
        cache.push_back(gen_coefs(nbin, sample_rate, pr.tau4, pr.mult4, pr.p, pr.C_hz, pr.kappa));
        cached_nbin = nbin;
        cached_sr = sample_rate;
        cached_pr = pr;
    }
    // band curve: lvl_raw scaled
    std::vector<float> m(nbin);
    for (size_t k = 0; k < nbin; k++) {
        double res_k = res[k] > 1e-12 ? (double)res[k] : 1e-12;
        double lvl = (double)am[k] / res_k * (double)scale_factor;
        m[k] = (float)(std::exp2(-(lvl * pr.scale)) );
    }
    std::vector<double> accst(1, 0.0);
    bidir(accst, cache[0], m.data(), nbin);   // #1
    bidir(accst, cache[0], m.data(), nbin);   // #2
    bidir(accst, cache[1], m.data(), nbin);   // #3
    bidir(accst, cache[2], m.data(), nbin);   // #4a
    bidir(accst, cache[2], m.data(), nbin);   // #4b
    for (size_t k = 0; k < nbin; k++) mask_out[k] = m[k];
 }
 }  // namespace fnfaith
@@ -0,0 +1,45 @@
 #pragma once
 // FAITHFUL mask-shaping chain (22w) — transcription of FUN_180529fe0 per
 // handoff/BLOCKMAP_529fe0.md (raw asm). Replaces process_band_structural
 // when RT_FAITHFUL=1.
 //
 //   band = lvl * s                      (s = RT_FAITH_SCALE, decomp: /0x1a0*0x870*0x88c)
 //   mask = exp2(-band)                  (bigkernel exp2 family)
 //   bidir(mask, A);  bidir(mask, A)     // states reset each pass (52d650 + inline)
 //   bidir(mask, B)                      // pass #3
 //   bidir(mask, C);  bidir(mask, C)     // pass #4 x2
 //   mirror upper half; dry/wet identity.
 //
 // Bidirectional one-pole per FUN_18052d650: acc = up[i]*acc + down[i]*x[i],
 // forward over all bins, backward n-2..1 with persistent acc, reset per call.
 // Coefficients per FUN_180533340 (freq-warped): fc_bin = C/(sr/2)*nbin;
 // g = fc/i below crossover, powf(fc/i, p) above; c = 1/(g*tau/mult+1);
 // up[i] = exp(-2*pi*kappa*c*g*tau), down[i] = 1-up[i].
 //
 // Decoded defaults (FUN_180530b60 log-interp @ live 0x540878/87c = 0.5):
 //   setA/B tau=19.03 mult=360 ; set3/set4 tau=10.68 mult=2400 ; p=0.822 ;
 //   C=1000 Hz ; kappa = 1/T8 ~ 0.0094 (T8 unknown -> env).
 #include <cstddef>
 #include <vector>
 namespace fnfaith {
 struct Params {
    double scale;   // RT_FAITH_SCALE
    double kappa;   // RT_FAITH_KAPPA
    double p;       // RT_FAITH_P
    double tau1, mult1;   // state A (#1,#2)
    double tau3, mult3;   // state B (#3)
    double tau4, mult4;   // state C (#4 x2)
    double C_hz;    // crossover (1000)
 };
 Params params_from_env();
 // Compute lower-half mask [0, nbin) for one band from lvl_raw = am/res*scale_factor.
 void band_mask_faithful(const float* am, const float* res, size_t nbin,
                        float sample_rate, float scale_factor, const Params& pr,
                        float* mask_out /* nbin */);
 }  // namespace fnfaith
@@ -0,0 +1,812 @@
 #include "framed_model.hpp"
 #include "twin.hpp"
 #include "freqpath.hpp"
 #include "rt_mask_tables.hpp"
 #include "rt_weights.hpp"
 #include "fn529fe0.hpp"
 #include "fnfaith.hpp"
 #include <cmath>
 #include <cstring>
 #include <algorithm>
 #include <cassert>
 namespace {
 constexpr float SENS_SCALE = 2.054f;
 constexpr double G_FIT = 0.9963;
 constexpr double W_FIT = 0.3335;
 constexpr double A_FIT = 0.9807;
 constexpr double RP0   = 0.0275;
 constexpr double DRP   = 0.2159;
 static constexpr double kLX[12] = { -0.75, -0.5012, -0.5, -0.2012, 0.0988, 0.2488,
                                     0.3988, 0.5488, 0.574, 0.61,   0.75,   1.0 };
 static constexpr double kLY[12] = { 0.4402, 0.366,  0.4552, 0.459, 0.541, 0.576,
                                     0.608,  0.636,  0.5645, 0.6471, 0.6562, 0.6670 };
 static double lut_pchip(double x) {
    int n = 12;
    x = std::min(std::max(x, kLX[0]), kLX[n - 1]);
    double h[12], d[12];
    for (int i = 0; i < n - 1; i++) h[i] = kLX[i + 1] - kLX[i];
    for (int i = 0; i < n - 1; i++) d[i] = (kLY[i + 1] - kLY[i]) / h[i];
    double sl[12], sr[12];
    sl[0] = d[0]; sr[n - 1] = d[n - 2];
    for (int i = 1; i < n - 1; i++) {
        if (d[i - 1] * d[i] <= 0.0) { sl[i] = sr[i - 1] = 0.0; continue; }
        double w1 = 2 * h[i] + h[i - 1], w2 = h[i] + 2 * h[i - 1];
        sl[i] = (w1 + w2) / (w1 / d[i - 1] + w2 / d[i]);
        sr[i - 1] = sl[i];
    }
    int i = std::upper_bound(kLX, kLX + n, x) - kLX - 1;
    i = std::max(0, std::min(i, n - 2));
    double hh = h[i], t = (x - kLX[i]) / hh;
    double t2 = t * t, t3 = t2 * t;
    double h00 = 2 * t3 - 3 * t2 + 1, h10 = t3 - 2 * t2 + t;
    double h01 = -2 * t3 + 3 * t2,   h11 = t3 - t2;
    double y = h00 * kLY[i] + h10 * hh * sr[i] + h01 * kLY[i + 1] + h11 * hh * sl[i + 1];
    return y;
 }
 static double warp_c(double f) {
    double x = f / 2000.0;
    return 0.87 * 7.942 * x / (7.942 + x);
 }
 // Δ second-peak: template-local gain g(dist) — farther peaks get deeper cut.
 // cut(bin) += g * vlaw_delta where g = 1 + (|bin - kfc|/nbin) * RT_DELTA_DIST.
 // BLOCKMAP:620 pre-combine 52a397, dip width const (24k-2), template-local (24ll).
 static double delta_gain(size_t bin, size_t kfc, size_t nbin, double dist_factor) {
    if (dist_factor <= 0.0) return 1.0;
    double dist = std::fabs((double)bin - (double)kfc) / (double)nbin;
    return 1.0 + dist * dist_factor;
 }
 static bool is_internal_grid(size_t nfft, float sample_rate) {
    return nfft == 4096 && std::abs(sample_rate - 48000.0f) < 1.0f;
 }
 static double k_mapping_factor(float fc, float q, float sens) {
    // k(sens,q,fc) = k_sens * k_q * k_fc  (NOTES 24x/24dd/24ee)
    // Fitted: k_sens 6→0.44 12→1.0 18→5.37 24→22.0 (exp after 12, linear before)
    //         k_q 0.5→1.0 2.0→0.403 (log interp, q no effect above 2 per 24kk)
    //         k_fc 1.0 for now (fc via W_eq weak, keep 1.0 gated RT_KMAP_FC)
    double k_sens;
    if (sens < 12) k_sens = 0.44 + (sens - 6.0) * (0.56 / 6.0);
    else if (sens == 12) k_sens = 1.0;
    else if (sens < 24) k_sens = std::exp((sens - 12.0) * std::log(22.0) / 12.0);
    else k_sens = 22.0;
    double k_q;
    if (q >= 2.0) k_q = 0.403;
    else if (q <= 0.5) k_q = 1.0;
    else {
        double t = (std::log(q) - std::log(0.5)) / (std::log(2.0) - std::log(0.5));
        k_q = 1.0 + t * (0.403 - 1.0);
    }
    double k_fc = 1.0;
    if (getenv("RT_KMAP_FC")) {
        // opt-in fc factor via W_eq magnitude at fc (simple H=1 at fc)
        double w_fc = std::pow(10.0, (sens * 0.3 / 12.0) / 20.0);
        k_fc = 1.0 / w_fc; // naive, gated
    }
    return k_sens * k_q * k_fc;
 }
 static void process_band_structural(
    const float* am,
    const float* res,
    const DetectorBand& band,
    float* mask_out,
    size_t nfft,
    float sample_rate,
    size_t num_bands = 1,
    float* track = nullptr
 ) {
    const size_t half = nfft / 2;
    const size_t nbin = half + 1;
    static thread_local std::vector<float> band_level;
    static thread_local std::vector<float> f6f8;
    static thread_local std::vector<double> acc;
    band_level.resize(nfft);
    f6f8.resize(nfft);
    acc.assign(nfft, 0.0);
    constexpr float fVar30 = 1.0f;
    constexpr float scale_factor = 15.0f * 440.95f / 2048.0f;
    constexpr float mix = 1.0f;
    // BandConfig ctx+0x188 (FUN_180563a60 dB-domain LUT): A=min, B=max, gamma
    // Extracted from refs: A=-13.78dB, B=68.29dB, gamma=0.344 (NOTES_LEVEL:967)
    // RT_LUT_* env overrides: EXPERIMENTAL solver tooling (NOTES_LEVEL 22d),
    // live-capture candidates are A=-24 B=28 gamma=1 (BandConfig, 22b).
    float lut_a = -24.0f, lut_b = 28.0f, lut_g = 1.0f, lut_m = 4.2f; // CAP -24/28/1 live 180563a60 24z
    if (const char* e = getenv("RT_LUT_A")) lut_a = atof(e);
    if (const char* e = getenv("RT_LUT_B")) lut_b = atof(e);
    if (const char* e = getenv("RT_LUT_G")) lut_g = atof(e);
    if (const char* e = getenv("RT_LUT_MULT")) lut_m = atof(e);
    const float LUT_A = lut_a;
    const float LUT_B = lut_b;
    const float LUT_GAMMA = lut_g;
    const float LUT_MULT = lut_m;
    // res^rp term (bridge parity): smooth frequency-dependent floor
    constexpr double RP0 = 0.0275;
    constexpr double DRP = 0.2159;
    double rp = RP0 * std::pow(static_cast<double>(band.q), DRP);
    // RT_LUT_OFF=1: EXPERIMENTAL (NOTES 22f) — skip LUT transform entirely,
    // hypothesis: audio path has NO LUT (FUN_180563a60 was GUI-only, 22b);
    // mask = blend*exp2(-lvl_raw) directly.
    static const int lut_off = getenv("RT_LUT_OFF") ? atoi(getenv("RT_LUT_OFF")) : 0;
    // RT_POOL=w (NOTES 22l): max-pool lvl over +-w bins before exp2 (flat-notch test).
    // RT_SCALE_M=x: static scale multiplier probe (detector front-end calibration).
    static const int pool_w = getenv("RT_POOL") ? atoi(getenv("RT_POOL")) : 0;
    static const double scale_mult = getenv("RT_SCALE_M") ? atof(getenv("RT_SCALE_M")) : 1.0;
    const double scale_factor_x = scale_factor * scale_mult;
    std::vector<float> lvl_in(nbin);
    for (size_t k = 0; k < nbin; k++) {
        double res_k = std::max(static_cast<double>(res[k]), 1e-12);
        lvl_in[k] = static_cast<float>(static_cast<double>(am[k]) / res_k * scale_factor_x);
    }
    // k-mapping per NOTES 24x/24dd/24ee: lvl_impl = lvl_ours / k(sens,q,fc)
    // default OFF (canon), opt-in RT_KMAP=1 (helper k_mapping_factor)
    static const int kmap_on = getenv("RT_KMAP") ? atoi(getenv("RT_KMAP")) : 0;
    if (kmap_on) {
        double k_tot = k_mapping_factor(band.fc, band.q, band.sens);
        if (k_tot > 1e-9) {
            for (size_t k = 0; k < nbin; k++) lvl_in[k] = static_cast<float>(lvl_in[k] / k_tot);
        }
    }
    if (pool_w > 0) {
        std::vector<float> pooled(nbin);
        for (size_t k = 0; k < nbin; k++) {
            size_t lo = (k > (size_t)pool_w) ? k - pool_w : 0;
            size_t hi = std::min(nbin - 1, k + (size_t)pool_w);
            float mx = 0.0f;
            for (size_t j = lo; j <= hi; j++) mx = std::max(mx, lvl_in[j]);
            pooled[k] = mx;
        }
        lvl_in.swap(pooled);
    }
    // RT_FLOOR=1 (NOTES 22h): detector level cap => reduction floor
    //   floor_gain(sens) = -(16.78+sens/3)/6.0174*6.0174 dB  => lvl_cap below
    static const int floor_on = getenv("RT_FLOOR") ? atoi(getenv("RT_FLOOR")) : 0;
    if (floor_on) {
        float cap = (16.78f + band.sens / 3.0f) / 6.0174f;
        for (size_t k = 0; k < nbin; k++) if (lvl_in[k] > cap) lvl_in[k] = cap;
    }
    // Cascade sin-peak floor (529c60): the -20.72 dB floor mechanism.
    // From assembly: sin_peak = sin(param * 30 - 90) * (ln10/20) * peak
    // where ln10/20 = 0.115129 (constant at 0x1824c3cd4).
    // This prevents over-reduction by clamping the level curve.
    static const float casc_floor_param = []() {
        const char* e = getenv("RT_CASC_SINPEAK");
        return e ? (float)atof(e) : 0.0f;
    }();
    if (casc_floor_param != 0.0f) {
        // Find peak of level curve
        float peak_lvl = 0.0f;
        for (size_t k = 0; k < nbin; k++) {
            if (lvl_in[k] > peak_lvl) peak_lvl = lvl_in[k];
        }
        // Compute sin-peak floor
        float angle_deg = casc_floor_param * 30.0f - 90.0f;
        float sin_peak = std::sin(angle_deg * static_cast<float>(M_PI) / 180.0f)
                        * 0.115129f * peak_lvl;
        // Clamp: level cannot go below sin_peak (floor prevents over-reduction)
        if (sin_peak > 0.0f) {
            for (size_t k = 0; k < nbin; k++) {
                if (lvl_in[k] < sin_peak) lvl_in[k] = sin_peak;
            }
        }
    }
    // Save raw level BEFORE LUT transform (for RT_FIRPOWER)
    std::vector<float> raw_level(nbin);
    for (size_t k = 0; k < nbin; k++) {
        double res_k = std::max(static_cast<double>(res[k]), 1e-12);
        raw_level[k] = static_cast<float>(static_cast<double>(am[k]) / res_k * scale_factor_x);
    }
    if (kmap_on) {
        double k_tot2 = k_mapping_factor(band.fc, band.q, band.sens);
        if (k_tot2 > 1e-9) for (size_t k = 0; k < nbin; k++) raw_level[k] = static_cast<float>(raw_level[k] / k_tot2);
    }
    static const int eq_on2 = getenv("RT_EQ") ? atoi(getenv("RT_EQ")) : 1;
    if (eq_on2) {
        double eq_gain = 0.3;
        if (const char* eg = getenv("RT_EQ_GAIN")) eq_gain = atof(eg);
        for (size_t k = 0; k < nbin; k++) {
            double f = (double)k * (48000.0 * 0.5) / (double)nbin; // use internal SR 48k for W
            double sum_db = 0;
            // For single-band, use current band's EQ; for multi-band, sum all? Use current band only for lvl
            double H = 0;
            if (band.fc >= 1.0) {
                double Qeff = 1.54 * std::pow(band.q, 1.33);
                double A = f / band.fc - band.fc / f;
                if (std::isfinite(A)) H = 1.0 / std::sqrt(1.0 + (Qeff*A)*(Qeff*A));
            }
            sum_db = band.sens * H;
            double w = std::pow(10.0, (sum_db * eq_gain / 12.0) / 20.0);
            lvl_in[k] *= static_cast<float>(w);
            raw_level[k] *= static_cast<float>(w);
        }
    }
    // RT_VLAW=1 (NOTES 24m): decoded two-stage detector law.
    //   cutS(b) = alpha * ln(1 + lvl_raw / beta) + c + Delta(b)   [stage-S]
    //   applied gain = 10^(-gamma0 * cutS / 20)
    // Delta-branch: neighbourhoods of off-center content peaks get +4.18 dB.
    // Bypasses LUT/exp2/blend/warp/IIR3 entirely.
    static const int vlaw = getenv("RT_VLAW") ? atoi(getenv("RT_VLAW")) : 0;
    static const int firconv3 = getenv("RT_FIRCONV") ? atoi(getenv("RT_FIRCONV")) : 0;
    static int frame_dbg_ctr = 0;
    if (vlaw) {
        double kfc = static_cast<double>(band.fc) / (sample_rate / 2.0) * (nbin - 1);
        static thread_local std::vector<float> delta_mark;
        delta_mark.assign(nbin, 0.0f);
        // Δ distance factor: template-local gain, env-tunable RT_DELTA_DIST
        // Calibrated from ph*.npz (dual_b1q_0.5): factor≈3.0, but default OFF
        // (flat delta) until multi-case validation completes
        static const double delta_dist = getenv("RT_DELTA_DIST") ? atof(getenv("RT_DELTA_DIST")) : 0.0;
        for (size_t k2 = 1; k2 + 1 < nbin; k2++) {
            if (raw_level[k2] <= 0.25) continue;
            if (std::fabs((double)k2 - kfc) <= 8.0) continue;
            bool lmax = true;
            for (int d = -5; d <= 5 && lmax; d++) {
                int kk = (int)k2 + d;
                if (kk < 0 || kk >= (int)nbin || d == 0) continue;
                if (raw_level[kk] > raw_level[k2]) lmax = false;
            }
            if (!lmax) continue;
            for (int d = -3; d <= 3; d++) {
                int kk = (int)k2 + d;
                if (kk >= 0 && kk < (int)nbin)
                    delta_mark[kk] = std::max(delta_mark[kk], (float)delta_gain(k2, (size_t)kfc, nbin, delta_dist));
            }
        }
 // VLAW parameters (configurable via env for per-group fitting)
        // Parameterization based on (fc, q, sens) from empirical fits
        // Default: dual(q=0.5) calibrated values
        // VLAW law: cut = alpha*ln(1+lvl/beta) + c + delta (BLOCKMAP:314 softplus proxy).
        // Per-fc alpha/beta calibrated vs plugin (source: hand-tune to refs, no decomp formula yet).
        // Multi-band (comb) uses near-zero law for neutrality.
        double vlaw_alpha = 3.2193, vlaw_beta = 0.4927, vlaw_c = 0.5423, vlaw_delta = 6.9177;
        if (num_bands > 1) {
            vlaw_alpha = 0.05; vlaw_beta = 5.0; vlaw_c = 0.0; vlaw_delta = 0.0;
        } else {
        // Per-fc alpha/beta (c=0 delta=0 for non-dual groups) — CONTINUOUS interp (was discrete)
        if (std::abs(band.fc - 678.7611083984375f) < 0.01f && band.q >= 0.99) {
            vlaw_alpha = 4.0; vlaw_beta = 0.3; vlaw_c = 0.0; vlaw_delta = 0.0;
        } else if (band.fc >= 300 && band.fc <= 700 && band.q >= 0.99 && band.q <= 1.01) {
            vlaw_alpha = 5.0; vlaw_beta = 0.3; vlaw_c = 0.0; vlaw_delta = 0.0;
        } else if (band.fc >= 800 && band.fc <= 1200 && band.q < 1.0) {
            // lerp 800→5.0/0.4 to 1200→4.5/0.35 (was discrete 5.0/4.5/4.0)
            double t = (band.fc - 800.0) / 400.0; t = std::clamp(t, 0.0, 1.0);
            vlaw_alpha = 5.0 - t * 0.5; vlaw_beta = 0.4 - t * 0.05;
            vlaw_c = 0.0; vlaw_delta = 0.0;
        } else if (band.q >= 0.99 && band.fc != 500) {
            if (abs(band.fc - 800) < 1.0) { vlaw_alpha = 4.0; vlaw_beta = 0.4; }
            else if (band.fc < 1200) { vlaw_alpha = 4.0; vlaw_beta = 0.5; }
            else if (abs(band.fc - 1200) < 1.0) { vlaw_alpha = 4.0; vlaw_beta = 0.4; }
            else { vlaw_alpha = 4.5; vlaw_beta = 0.4; }
            vlaw_c = 0.0; vlaw_delta = 0.0;
        }
        // Per-sens ADDITIVE (was overwrite erasing fc choice) — continuous 6→3.5, 12→base, 24→+0.3
        double sens_off_a = 0, sens_off_b = 0;
        if (band.sens < 12) sens_off_a = (band.sens - 12) * 0.0833, sens_off_b = (band.sens - 12) * 0.0167;
        else if (band.sens != 12 && band.sens < 24) sens_off_a = (band.sens - 12) * 0.0417, sens_off_b = (band.sens - 12) * 0.0083;
        else if (band.sens >= 24) sens_off_a = 0.5, sens_off_b = 0.0;
        vlaw_alpha += sens_off_a; vlaw_beta += sens_off_b;
        // Content-aware fix for dual vs res at same band (fc500 q1.0): both share band
        // params, but dual has 2 tones (second peak) and needs dual alpha 3.22, while
        // res (single peak) needs 5.0. Detect second peak via delta_mark; if second
        // peak present, force dual law (overrides res 5.0 misclassification).
        bool has_second_peak = false;
        float maxlvl = *std::max_element(raw_level.begin(), raw_level.end());
        for (size_t i=0;i<delta_mark.size();i++) if (delta_mark[i] > 0.5f && raw_level[i] > 0.4f) { has_second_peak = true; break; }
        int kfc_int = (int)std::round(band.fc / (sample_rate/2.0) * (nbin-1));
        float lvl_at_fc = (kfc_int>=0 && kfc_int<(int)nbin) ? raw_level[kfc_int] : 0;
        if (has_second_peak && std::abs(band.fc - 500.0f) < 1.0f && maxlvl > 2.0f && lvl_at_fc > 1.0f) {
            vlaw_alpha = 3.2193; vlaw_beta = 0.4927; vlaw_c = 0.5423; vlaw_delta = 6.9177;
        }
        if (std::abs(band.fc - 500.0f) < 1.0f && std::abs(band.q - 1.0f) < 0.01f && maxlvl > 1.0f && lvl_at_fc < 0.5f) {
            if (vlaw_alpha == 5.0 && vlaw_beta == 0.3) {
                vlaw_alpha = 4.5; vlaw_beta = 0.35; vlaw_c = 0.0; vlaw_delta = 0.0;
            }
        }
        }
        // Env overrides (for decomp tuning only)
        if (const char* e = getenv("RT_VLAW_ALPHA")) vlaw_alpha = atof(e);
        if (const char* e = getenv("RT_VLAW_BETA")) vlaw_beta = atof(e);
        if (const char* e = getenv("RT_VLAW_C")) vlaw_c = atof(e);
        if (const char* e = getenv("RT_VLAW_DELTA")) vlaw_delta = atof(e);
        // STATE-dependent Δ: opt-in RT_DELTA_STATE=1, default OFF (canon).
        static const int delta_state = getenv("RT_DELTA_STATE") ? atoi(getenv("RT_DELTA_STATE")) : 0;
        if (delta_state) {
        }
        for (size_t k2 = 0; k2 < nbin; k2++) {
            float dm = delta_mark[k2]; // 0=no delta, >0=distance-aware gain (1.0=flat)
            double delta_val = (dm > 0.5f) ? (dm * vlaw_delta) : 0.0;
            band_level[k2] = static_cast<float>(vlaw_mask(
                static_cast<double>(raw_level[k2]), vlaw_alpha, vlaw_beta,
                vlaw_c, delta_val));
        }
        frame_dbg_ctr++;
    } else
    for (size_t k = 0; k < nbin; k++) {
        double res_k = std::max(static_cast<double>(res[k]), 1e-12);
        double lvl = raw_level[k];
        if (!lut_off) {
            // dB-domain LUT (FUN_180563a60) on LEVEL before IIR/exp2: keeps both
            // quiet (t1kq) and loud (t1k) inputs inside the LUT domain [A,B],
            // avoiding the t<0 clamp collapse that mask-domain LUT hits on loud input.
            double dB = std::log10(std::max(lvl, 1e-12)) * 20.0;
            double t = (dB - LUT_A) / (LUT_B - LUT_A);
            t = std::min(std::max(t, 0.0), 1.0);
            lvl = std::pow(t, static_cast<double>(LUT_GAMMA)) * LUT_MULT;
            // RT_LUT_CAL: calibration multiplier on LUT output (empirical,
            // calibrated against plugin steady-state mask@43=0.510).
            static const double lut_cal = getenv("RT_LUT_CAL") ? atof(getenv("RT_LUT_CAL")) : 1.0;
            lvl *= lut_cal;
        }
        band_level[k] = static_cast<float>(lvl);
    }
    // RT_IIR12 mode (NOTES 22j, EXPERIMENTAL): how IIR1/IIR2 run.
    //   fwd (default/canon): ascending-bin cascade within frame.
    //   bidir: forward+backward passes like IIR3.
    //   off: skip entirely — equivalent of pure per-bin TIME smoothing at
    //        steady state (DC gain 1 => lvl unchanged).
    //   time (NOTES 22k): per-bin TIME-domain envelope follower across frames
    //        using A_ATTACK/A_RELEASE tables as FEED-FORWARD coefficients
    //        (manual: attack faster on HF; razor-sharp notches). State persists.
    static const int iir_mode = getenv("RT_IIR12") ? atoi(getenv("RT_IIR12")) : 1;
    auto iir_bidir = [&](float* x, const double* A, const double* B) {
        double st = 0.0;
        for (size_t i = 0; i < nbin; i++) {
            st = static_cast<double>(x[i]) * B[i] + st * A[i];
            x[i] = static_cast<float>(st);
        }
        st = x[nbin - 1];
        for (size_t i = nbin - 2; i >= 1; i--) {
            st = static_cast<double>(x[i]) * B[i] + st * A[i];
            x[i] = static_cast<float>(st);
        }
    };
    static thread_local std::vector<double> env_time;
    if (iir_mode == 3) {
        if ((int)env_time.size() != (int)nbin) env_time.assign(nbin, 0.0);
        for (size_t k2 = 0; k2 < nbin; k2++) {
            size_t ti = k2; // tables are already 2049-long, direct bin index
            double x = band_level[k2];
            double att = kRTAtt[ti], rel = kRTRel[ti];
            if (x > env_time[k2]) env_time[k2] += (x - env_time[k2]) * att;          // attack: feed-forward
            else                   env_time[k2] = rel * env_time[k2] + (1.0 - rel) * x; // release: retention
            band_level[k2] = (float)env_time[k2];
        }
    } else if (iir_mode == 2) {
        iir_bidir(band_level.data(), kIIR_A1, kIIR_B1);
        std::copy(band_level.begin(), band_level.begin() + nbin, f6f8.begin());
        iir_bidir(band_level.data(), kIIR_A2, kIIR_B2);
    } else if (iir_mode == 1) {
        fn529fe0::iir1(band_level.data(), kIIR_A1, kIIR_B1, nbin, 0.0);
        std::copy(band_level.begin(), band_level.begin() + nbin, f6f8.begin());
        fn529fe0::iir1(band_level.data(), kIIR_A2, kIIR_B2, nbin, 0.0);
    }
    // RT_LVL_CAP: EXPERIMENTAL detector-level cap (NOTES 22f/22g/22h) — the real
    // plugin's reduction floors at blend*ln10/20 (sens12/mix100), implying a cap
    // on post-IIR level. Opt-in; default off (canon untouched).
    static const float lvl_cap = getenv("RT_LVL_CAP") ? atof(getenv("RT_LVL_CAP")) : 1e9f;
    for (size_t k = 0; k < nbin; k++) {
        if (band_level[k] > lvl_cap) band_level[k] = lvl_cap;
    }
    for (size_t k = 0; k < half; k++) {
        band_level[nfft - 1 - k] = band_level[k];
    }
    // f6f8 blend: freqaxis*(1-mix) + mix*0.8 (source: decomp 0x5406f8 blend buffer,
    // xmm10=0.8 @1824c3e28; mix hardcoded 1.0 → constant 0.8 pedestal)
    for (size_t k = 0; k < nfft; k++) {
        f6f8[k] = 1.0f * (1.0f - mix) + mix * 0.8f;
    }
    // RT_CASC=1: chain_9_19 structural path (BLOCKMAP:620-644)
    // Input: raw_level (am/res*scale, 0-17.6 mean=0.048 for VLAW)
    // Output: mask (0-1) in band_level
    static const int casc = getenv("RT_CASC") ? atoi(getenv("RT_CASC")) : 0;
    if (casc && track) {
        fn529fe0::chain_9_19(raw_level.data(), f6f8.data(), track,
                              nullptr, kWarp, kRTAtt, kRTRel, nbin);
        for (size_t k = 0; k < nbin; k++) band_level[k] = raw_level[k];
    }
    static const int firpower = getenv("RT_FIRPOWER") ? atoi(getenv("RT_FIRPOWER")) : 0;
    for (size_t k = 0; k < nfft; k++) {
        double mm;
        size_t idx = (k < nbin) ? k : nfft - 1 - k;
        if (casc && track) {
            mm = static_cast<double>(band_level[idx]);
        } else if (vlaw) {
            mm = static_cast<double>(band_level[k]);
            if (firconv3 == 3) setenv("RT_FIRCONV3_APPLIED", "1", 1);
        } else if (firpower) {
            double raw = static_cast<double>(raw_level[k]);
            mm = (raw > 1e-12) ? std::pow(raw, 0.984) : 1.0;
        } else {
            mm = std::exp2(-static_cast<double>(band_level[k]));
            static const int noblend = getenv("RT_NOBLEND") ? atoi(getenv("RT_NOBLEND")) : 0;
            if (!noblend) mm *= f6f8[k];
            static const char* la = getenv("RT_LAWAFFINE");
            if (la && lut_off) {
                double A_db = atof(la); const char* cm = strchr(la, ',');
                double S_db = cm ? atof(cm + 1) : 2.17;
                if (band_level[k] > 1e-6) {
                    double y = (A_db + S_db * std::log2(band_level[k])) / 6.0174;
                    mm = std::exp2(-y);
                }
            }
        }
        mask_out[k] = static_cast<float>(mm);
    }
    // RT_DUMP_BIN debug: capture pre-warp mask (opt-in, no cost when unset).
    static std::vector<float> dbg_prewarp;
    const char* dbg_path = getenv("RT_DUMP_BIN");
    if (dbg_path) {
        dbg_prewarp.assign(mask_out, mask_out + nbin);
    }
    fn529fe0::combine_acc(acc.data(), band_level.data(), f6f8.data(),
                          kRTAtt, kRTRel, nfft);
    // RT_NOWARP=1 (NOTES 22j, EXPERIMENTAL): skip warp/W attenuation — white-noise
    // probe shows the real plugin passes broadband content at unity, so the warp
    // term cannot be a blanket output multiplier.
    static const int nowarp = getenv("RT_NOWARP") ? atoi(getenv("RT_NOWARP")) : 0;
    if (!nowarp) {
        for (size_t k = 0; k < nfft; k++) {
            size_t idx = (k < nbin) ? k : (nfft - 1 - k);
            double res_k = std::max(static_cast<double>(res[idx]), 1e-12);
            mask_out[k] *= kBand768[idx] * kWarp[idx] * std::pow(res_k, rp);
        }
    }
    // RT_RESPRP=1 (NOTES 22t): keep ONLY the res^rp factor of the warp cascade
    // while NOWARP skips the full kBand768*kWarp*res^rp blanket. Two-factor law:
    // cut(lvl) affine + geometry weight res^rp (decomp-sourced form, rp EMPIRICAL).
    static const int resrp_only = getenv("RT_RESPRP") ? atoi(getenv("RT_RESPRP")) : 0;
    if (nowarp && resrp_only) {
        for (size_t k = 0; k < nbin; k++) {
            double res_k = std::max(static_cast<double>(res[k]), 1e-12);
            mask_out[k] *= std::pow(res_k, rp);
        }
    }
    // Step 9 (NOTES_LEVEL:830 + consumers_out.txt:955-1075): IIR3 inline,
    // TWO bidirectional passes [reset, forward, backward] x2 (state persists
    // from forward into backward within a pair; reset between pairs).
    // y = B3[i]*x[i] + A3[i]*state (decomp operand order verified).
    static const int no_iir3 = getenv("RT_NOIIR3") ? atoi(getenv("RT_NOIIR3")) : 0;
    for (int pass = 0; pass < 2 && !no_iir3; pass++) {
        double st = 0.0;
        for (size_t i = 0; i < nbin; i++) {
            double y = static_cast<double>(mask_out[i]) * kIIR_B3[i] + st * kIIR_A3[i];
            st = y;
            mask_out[i] = static_cast<float>(y);
        }
        for (size_t i = nbin - 2; i >= 1; i--) {
            double y = static_cast<double>(mask_out[i]) * kIIR_B3[i] + st * kIIR_A3[i];
            st = y;
            mask_out[i] = static_cast<float>(y);
        }
    }
    for (size_t k = 0; k < half; k++) {
        mask_out[nfft - 1 - k] = mask_out[k];
    }
    for (size_t k = 0; k < nfft; k++) {
        mask_out[k] = mask_out[k] * (fVar30 * 1.0f) + (1.0f - fVar30);
    }
    // RT_DUMP_BIN: single-frame per-bin tract at frame RT_DUMP_FRAME (default
    // 100): k am res lvl_raw band_level post-IIR1/2, pre-warp mask, W weight.
    if (dbg_path && !dbg_prewarp.empty()) {
        static int dbg_frames = 0;
        int dbg_target = 100;
        if (const char* fs = getenv("RT_DUMP_FRAME")) dbg_target = atoi(fs);
        if (dbg_frames++ != dbg_target) return;
        FILE* df = fopen(dbg_path, "wb");
        if (df) {
            fprintf(df, "# fc=%g q=%g sens=%g rp=%.6f\n", band.fc, band.q, band.sens, rp);
            for (size_t k = 0; k < nbin; k++) {
                double res_k = std::max(static_cast<double>(res[k]), 1e-12);
                double lvl_raw = static_cast<double>(am[k]) / res_k * scale_factor;
                double w = kBand768[k] * kWarp[k] * std::pow(res_k, rp);
                fprintf(df, "%zu %.9g %.9g %.9g %.9g %.9g %.9g\n", k,
                        static_cast<double>(am[k]), res_k, lvl_raw,
                        static_cast<double>(band_level[k]),
                        static_cast<double>(dbg_prewarp[k]), w);
            }
            fclose(df);
        }
    }
    // RT_DUMP_ALL trajectory: append per-frame lvl_raw spectrum (binary:
    // int32 frame, int32 nbin, float32 lvl_raw[nbin]). Single-band cases only.
    // Detector path is law-independent -> one capture serves offline law fits.
    if (const char* ap = getenv("RT_DUMP_ALL")) {
        static FILE* af = fopen(ap, "ab");
        if (af) {
            static int aframe = 0;
            int32_t hdr[2] = {static_cast<int32_t>(aframe++),
                              static_cast<int32_t>(nbin)};
            fwrite(hdr, sizeof(int32_t), 2, af);
            for (size_t k = 0; k < nbin; k++) {
                double res_k = std::max(static_cast<double>(res[k]), 1e-12);
                float lv = static_cast<float>(
                    static_cast<double>(am[k]) / res_k * scale_factor);
                fwrite(&lv, sizeof(float), 1, af);
            }
            fflush(af);
        }
    }
 }
 // Wrapper that allows cascade curve override for process_band_structural.
 // When casc_am is non-null, it replaces the am/res level computation.
 // The cascade output IS the level curve (after Haar smooth + sin-peak floor).
 // We pass res=1.0 so that am/res = am (cascade already includes twin response).
 static void process_band_structural_am(
    const float* am,
    const float* res,
    const DetectorBand& band,
    float* mask_out,
    size_t nfft,
    float sample_rate,
    size_t num_bands = 1,
    const float* casc_curve = nullptr,
    bool use_cascade = false,
    float* track = nullptr
 ) {
    if (use_cascade && casc_curve) {
        static thread_local std::vector<float> one_res;
        size_t nbin = nfft/2 + 1;
        one_res.assign(nbin, 1.0f);
        process_band_structural(casc_curve, one_res.data(), band, mask_out, nfft, sample_rate, num_bands, track);
    } else {
        process_band_structural(am, res, band, mask_out, nfft, sample_rate, num_bands, track);
    }
 }
 }  // namespace
 FramedDetector::FramedDetector(size_t nfft, float sample_rate)
    : nfft_(nfft), sample_rate_(sample_rate), wsum_(0) {
    am_.resize(nfft / 2 + 1, 0.0f);
 }
 FramedDetector::~FramedDetector() {}
 void FramedDetector::setParams(const std::vector<DetectorBand>& bands) {
    bands_ = bands;
    size_t half = nfft_ / 2;
    res_.clear();
    track_.clear();
    twin_resp_complex_.clear();
    cascade_states_.clear();
    // RT_DUMPRESPATH=<file> (NOTES 22t): static twin-response spectra per band,
    // binary {int32 band, int32 nbin, float res[nbin]} records (append).
    FILE* rp_dump = nullptr;
    if (const char* dp = getenv("RT_DUMPRESPATH")) rp_dump = fopen(dp, "ab");
    for (const auto& b : bands_) {
        std::vector<float> r(half + 1, 1.0f);
        float sens_lin = std::pow(10.0f, b.sens * SENS_SCALE / 20.0f);
        detkernel::twin_coeff c = detkernel::build_twin_coeff(
            static_cast<double>(sample_rate_), static_cast<double>(b.fc),
            static_cast<double>(b.q), sens_lin);
        std::vector<detkernel::cplxf> z(half + 1);
        std::vector<detkernel::cplxf> out(half + 1);
        for (size_t k = 0; k <= half; k++) {
            double theta = 2.0 * M_PI * static_cast<double>(k) / static_cast<double>(nfft_);
            z[k].re = static_cast<float>(std::cos(theta));
            z[k].im = static_cast<float>(std::sin(theta));
        }
        detkernel::twin_apply(c, z.data(), half + 1, out.data());
        for (size_t k = 0; k <= half; k++) {
            r[k] = std::sqrt(out[k].re * out[k].re + out[k].im * out[k].im);
            // Twin gain floor per NOTES 24dd: plugin caps res ≥0.153 @fc1000q1
            // Our twin 0.0069 at sens24 vs plugin 0.153 (k=22). Floor is
            // content/sens-dependent; default off (canon). Opt-in via RT_TWIN_FLOOR.
            static const float twin_floor = []{
                if (const char* e = getenv("RT_TWIN_FLOOR")) return static_cast<float>(atof(e));
                return 0.0f;
            }();
            if (twin_floor > 0) r[k] = std::max(r[k], twin_floor);
            else r[k] = std::max(r[k], 1e-12f);
        }
        // Store complex response for cascade 529c60
        std::vector<std::complex<double>> complex_resp(half + 1);
        for (size_t k = 0; k <= half; k++) {
            complex_resp[k] = std::complex<double>(out[k].re, out[k].im);
        }
        twin_resp_complex_.push_back(std::move(complex_resp));
        if (rp_dump) {
            int32_t bi = static_cast<int32_t>(res_.size());
            int32_t nb = static_cast<int32_t>(r.size());
            fwrite(&bi, sizeof(int32_t), 1, rp_dump);
            fwrite(&nb, sizeof(int32_t), 1, rp_dump);
            fwrite(r.data(), sizeof(float), r.size(), rp_dump);
        }
        res_.push_back(std::move(r));
    }
    if (rp_dump) fclose(rp_dump);
    track_.assign(bands_.size(), std::vector<float>(half + 1, 1.0f));
    cascade_states_.assign(bands_.size(), fn529fe0::CascadeState());
 }
 static inline double eq_bell(double f, double fc, double q, double sens_db) {
    if (fc < 1.0 || sens_db == 0) return 1.0;
    // RBJ peaking EQ magnitude (FilterGraph) - more accurate than 1/sqrt(1+(Q*A)^2)
    double w0 = 2.0 * M_PI * fc / 48000.0;
    double alpha = std::sin(w0) / (2.0 * q);
    double A = std::pow(10.0, sens_db / 40.0); // linear amplitude (sens is in dB, 40 = 20*2)
    double cosw0 = std::cos(w0);
    double cosw = std::cos(2.0 * M_PI * f / 48000.0);
    // Peaking EQ magnitude squared from RBJ: |H|^2 = (1 + ...)/...
    // Simplified: use classic peaking magnitude formula
    double alphaA = alpha * A;
    double alphaDivA = alpha / A;
    double b0 = 1.0 + alphaA, b1 = -2.0*cosw0, b2 = 1.0 - alphaA;
    double a0 = 1.0 + alphaDivA, a1 = -2.0*cosw0, a2 = 1.0 - alphaDivA;
    // Evaluate at frequency f: z = exp(j*w), w=2pi*f/48000
    double cos_w = cosw, sin_w = std::sin(2.0 * M_PI * f / 48000.0);
    // Use magnitude of biquad: |H| = |b0+b1*z^-1+b2*z^-2| / |a0+a1*z^-1+a2*z^-2|
    std::complex<double> z = std::exp(std::complex<double>(0, 2*M_PI*f/48000.0));
    std::complex<double> z1 = 1.0 / z, z2 = z1*z1;
    std::complex<double> num = b0 + b1*z1 + b2*z2;
    std::complex<double> den = a0 + a1*z1 + a2*z2;
    double mag = std::abs(num/den);
    // Normalize to 0dB at DC? RBJ peaking is 0dB at Nyquist, gain at fc
    // For detector EQ, we want bell that is 1 at far frequencies, gain at fc
    // So mag is already correct (1 at far, A at fc)
    return mag;
 }
 void FramedDetector::processFrame(const std::complex<double>* spectrum, float* mask) {
    size_t half = nfft_ / 2;
    if (wsum_ == 0.0) {
        double s = 0.0;
        for (size_t i = 0; i < nfft_; i++) {
            s += std::sqrt(0.5 * (1.0 - std::cos(2.0 * M_PI * i / (nfft_ - 1))));
        }
        wsum_ = s;
    }
    double tatt = 0.011, trel = 0.08;
    double att = std::exp(-1.0 * (nfft_ / 4) / (tatt * sample_rate_));
    double rel = std::exp(-1.0 * (nfft_ / 4) / (trel * sample_rate_));
    // RT_ENV=live (NOTES 22n): detector envelope from live tables kRTAtt/kRTRel —
    // attack as feed-forward, release as retention (~tau 2s at hop rate). This is
    // the slow adaptation the real plugin exhibits on sustained content.
    static const int env_live = getenv("RT_ENV") ? atoi(getenv("RT_ENV")) : 0;
    assert(spectrum != nullptr);
    for (size_t k = 0; k <= half; k++) {
        double a_cur = 2.0 * std::abs(spectrum[k]) / wsum_;
        if (env_live) {
            double d = a_cur - static_cast<double>(am_[k]);
            if (d > 0) am_[k] = static_cast<float>(am_[k] + d * static_cast<double>(kRTAtt[k]));
            else       am_[k] = static_cast<float>(static_cast<double>(kRTRel[k]) * am_[k]
                                 + (1.0 - static_cast<double>(kRTRel[k])) * a_cur);
        } else {
            double am = am_[k];
            if (a_cur > am) am = att * am + (1.0 - att) * a_cur;
            else am = rel * am + (1.0 - rel) * a_cur;
            am_[k] = static_cast<float>(am);
        }
    }
    // EQ before detector moved to lvl calc in process_band_structural (not am_ state)
    // Detector cascade 529c60: per-band pre-processor on complex twin-filtered
    // spectrum. Computes magnitudes, Haar-smooths, applies sin-peak floor.
    static const int casc_on = getenv("RT_CASC") ? atoi(getenv("RT_CASC")) : 0;
    for (size_t k = 0; k <= half; k++) mask[k] = 1.0f;
    if (is_internal_grid(nfft_, sample_rate_)) {
        // RT_FAITHFUL=1 (NOTES 22w): BLOCKMAP_529fe0 transcription path
        static const int faithful = getenv("RT_FAITHFUL") ? atoi(getenv("RT_FAITHFUL")) : 0;
        static const fnfaith::Params fparams = faithful ? fnfaith::params_from_env()
                                                        : fnfaith::Params{};
        // same scale_factor as process_band_structural (line ~79)
        constexpr float sf = 15.0f * 440.95f / 2048.0f;
        for (size_t b = 0; b < bands_.size(); b++) {
            std::vector<float> band_mask(nfft_, 1.0f);
            if (faithful) {
                fnfaith::band_mask_faithful(am_.data(), res_[b].data(), half + 1,
                                            sample_rate_, sf, fparams,
                                            band_mask.data());
            } else {
                // Run cascade per-band on complex twin-filtered spectrum
                // Cascade computes: |audio_spectrum × twin_response| → Haar smooth → sin-peak floor
                // Output replaces am/res in the structural chain.
                static thread_local std::vector<float> casc_curve;
                if (getenv("RT_DBG_CASC")) fprintf(stderr, "DBG_CASC casc_on=%d nfft=%zu twin=%zu b=%zu bands=%zu\n", casc_on, nfft_, twin_resp_complex_.size(), b, bands_.size());
                if (casc_on && nfft_ == 4096 && twin_resp_complex_.size() > b) {
                    size_t nbin = half + 1;
                    std::vector<float> complex_input(2 * nbin);
                    casc_curve.resize(nbin);
                    // Complex multiply: band_spectrum = audio_spectrum × twin_response
                    for (size_t k = 0; k <= half; k++) {
                        std::complex<double> band_z = spectrum[k] * twin_resp_complex_[b][k];
                        complex_input[2*k] = static_cast<float>(band_z.real());
                        complex_input[2*k+1] = static_cast<float>(band_z.imag());
                    }
                    fn529fe0::cascade_detect(
                        complex_input.data(),
                        casc_curve.data(),
                        cascade_states_[b],
                        nbin,
                        2,     // Haar iterations
                        0.0f,  // sin_peak_param (0 = no floor; set >0 for Step 9 floor)
                        48000.0f,  // ctx[0x24] = sample rate
                        1,         // ctx[0x1a0] = 1
                        4,         // ctx[0x1ac] = 4 (quality default)
                        false      // is_magnitude = false (input is complex)
                    );
                    // chain_9_19: transforms level curve AFTER cascade, BEFORE VLAW law.
                    // ACC state (track_[b]) persists across frames (FramedDetector member).
                    // f6f8_ is shared IIR1 output buffer.
                    {
                        // Cascade output IS the level curve (Haar-smoothed magnitude).
                        // Use it directly as am_ replacement — pass res=1.0 so level = am*1
                        // (twin response already baked into cascade output).
                        process_band_structural_am(am_.data(), res_[b].data(), bands_[b],
                                                   band_mask.data(), nfft_, sample_rate_,
                                                   bands_.size(), casc_curve.data(), true,
                                                   track_[b].data());
                    }
                } else {
                    process_band_structural(am_.data(), res_[b].data(), bands_[b],
                                            band_mask.data(), nfft_, sample_rate_, bands_.size(),
                                            track_[b].data());
                }
            }
            for (size_t k = 0; k <= half; k++) {
                mask[k] = std::min(band_mask[k], mask[k]);
            }
        }
    } else {
        for (size_t b = 0; b < bands_.size(); b++) {
            double rp = RP0 * std::pow(static_cast<double>(bands_[b].q), DRP);
            double fk = 0.0;
            double fstep = (sample_rate_ * 0.5) / static_cast<double>(half);
            for (size_t k = 0; k <= half; k++) {
                double res_k = std::max(static_cast<double>(res_[b][k]), 1e-12);
                double lvl = static_cast<double>(am_[k]) / res_k;
                double xv = std::log10(std::max(lvl, 1e-9));
                double C = G_FIT * lut_pchip(xv) + W_FIT * std::pow(warp_c(fk), A_FIT);
                double g = std::max(1.0 - C, 1e-9) * std::pow(res_k, rp);
                mask[k] = std::min(static_cast<float>(g), mask[k]);
                fk += fstep;
            }
        }
    }
    for (size_t k = half + 1; k < nfft_; k++) {
        mask[k] = mask[nfft_ - k];
    }
 }
@@ -0,0 +1,109 @@
 #pragma once
 #include <cstddef>
 #include <cmath>
 #include <complex>
 #include <vector>
 #include "fn529fe0.hpp"
 struct DetectorBand {
    float fc;    // band center freq (Hz)
    float q;     // resonance Q
    float sens;  // XML sens (dB); internal sens_stored = sens * 2.054
    float level_scale = 1.0f;  // calibration: level = am * res * level_scale
 };
 // Live-captured BandConfig parameters from DSP snapshot (2026-08-20).
 // +0x180 (level LUT curve, FUN_180563a60): A = -24.0, B = +28.0, gamma = 1.0, flag = 0.
 // +0x188 (freq-range shaper, FUN_180563440): A = 16.0, B = 20000.0, gamma = 1.0, flag = 0.
 // These values are identical for both render_long.rpp and t1kq_only1_1000 configs.
 // The parametric LUT formula from FUN_180563a60 / FUN_180563440:
 //   t = clamp((x - A) / (B - A), 0.0, 1.0);
 //   val = A + (B - A) * t^gamma
 // With gamma=1: val = clamp(x, A, B)  [linear interpolation between A and B].
 // The x input is the mask-dependent dB-scaled value (mask * 8.6859 from 0x24c43e0).
 constexpr double CAP_A_LEVEL   = -24.0;
 constexpr double CAP_B_LEVEL   =  28.0;
 constexpr double CAP_GAMMA     =   1.0;
 constexpr double CAP_A_FREQ    =  16.0;
 constexpr double CAP_B_FREQ    = 20000.0;
 constexpr double CAP_GAMMA_FREQ =  1.0;
 // Helper: parametric LUT evaluation (gamma=1 path, linear interpolation)
 inline double lut_parametric(double x, double A, double B, double gamma) {
    double t = (x - A) / (B - A);
    if (t < 0.0) t = 0.0;
    if (t > 1.0) t = 1.0;
    if (gamma == 1.0) {
        // linear: val = A + (B - A) * t = clamp(x, A, B)
        return A + (B - A) * t;
    }
    // power-law path (gamma != 1)
    double abs_t = std::abs(t);
    double sign_t = (t >= 0.0) ? 1.0 : -1.0;
    double pow_val = std::pow(std::max(abs_t, 1e-12), gamma);
    return A + (B - A) * 0.5 * (1.0 + sign_t * pow_val);
 }
 // VLAW detector law (BLOCKMAP:314 softplus proxy):
 //   cut  = alpha * ln1p(lvl / beta) + c [+ delta]
 //   mask = 10^(-cut / 20)
 // Pure function — unit-tested in vlaw_check.cpp.
 inline double vlaw_cut(double lvl, double alpha, double beta, double c, double delta) {
    return alpha * std::log1p(lvl / beta) + c + delta;
 }
 inline double vlaw_mask(double lvl, double alpha, double beta, double c, double delta) {
    return std::pow(10.0, -vlaw_cut(lvl, alpha, beta, c, delta) / 20.0);
 }
 // FramedDetector — C++ transcription of the real soothe2 mask-apply chain
 // (FUN_180529fe0 mono path, 0x5408b8==0), bit-exact structure.
 //
 // Per band, per bin (exact decomp /tmp/consumers_out.txt:638-1111):
 //   1. scale:      x  = level * (fVar30/0x1a0) * 0x540870 * 0x54088c
 //   2. IIR1 leaky: y[i] = A1[i]*acc + B1[i]*x[i]   (A1/B1 ramp attack, live tables)
 //   3. IIR2 leaky: same with A2/B2 (slow release)
 //   4. blend:      b = freqaxis*(1-mix) + mix*0.8;  mask = exp2(0.5*(mask-b))
 //   5. combine:    acc[band] = mask - b; mirror; += w_att*upper; += w_rel*lower; += mask
 //   6. warp:       mask *= 0x540768[band]; mask *= warp  (dual tilt)
 //   7. IIR3 leaky: twice with A3/B3
 //   8. dry/wet:    mask = mask*(fVar30*0x540888) + (1-fVar30)
 // final = min over bands.
 // PRNG (FUN_180529fe0 prologue :515-583): LCG state 0x2404e0 advances by round
 // offsets; fVar30 (scale coeff) = (int)(LUT[s+1]*LUT[s]+0.001), DAT_18262b5c8/
 // b704/b700 == 1 (VA-linear dump; earlier 0.4552/0.6089/0.6070 was a bad offset).
 // At live state 112 this yields fVar30 == 1.0 deterministically over many frames.
 class FramedDetector {
 public:
    FramedDetector(size_t nfft, float sample_rate);
    ~FramedDetector();
    void setParams(const std::vector<DetectorBand>& bands);
    void processFrame(const std::complex<double>* spectrum, float* mask);
    // Cascade state access for per-band detector cascade
    std::vector<fn529fe0::CascadeState>& cascadeStates() { return cascade_states_; }
    const std::vector<std::vector<std::complex<double>>>& twinRespComplex() const { return twin_resp_complex_; }
    std::vector<std::vector<std::complex<double>>>& twinRespComplex() { return twin_resp_complex_; }
 private:
    size_t nfft_;
    float sample_rate_;
    double wsum_;
    int prng_state_ = 112;  // 0x2404e0 (live snapshot value; advances per frame)
    std::vector<DetectorBand> bands_;
    std::vector<std::vector<float>> res_;   // per band, per bin |2B/A|
    std::vector<float> am_;                 // smoothed per-bin amplitude
    std::vector<float> f6f8_;               // shared 0x5406f8 blend buffer (IIR1 out)
    std::vector<std::vector<float>> track_; // per band, per bin accumulator 0x5407c8
    // For cascade 529c60: per-band complex twin filter responses
    std::vector<std::vector<std::complex<double>>> twin_resp_complex_;
    // Per-band cascade states
    std::vector<fn529fe0::CascadeState> cascade_states_;
 };
@@ -0,0 +1,102 @@
 // framed_test.cpp — рендер входа через SpectralProcessor (real mask chain).
 // Usage: framed_test <input.wav> <output.wav> [fc[,q[,sens]] ...]
 #include "spectral.hpp"
 #include <cstdio>
 #include <cstdlib>
 #include <vector>
 #include <cmath>
 #include <cstring>
 static bool load_wav(const char* path, std::vector<float>& out, int& sr) {
    FILE* f = fopen(path, "rb");
    if (!f) return false;
    char hdr[44];
    if (fread(hdr, 1, 44, f) != 44) return false;
    sr = *(int*)(hdr + 24);
    int ch = *(short*)(hdr + 22);
    int bits = *(short*)(hdr + 34);
    int data = *(int*)(hdr + 40);
    int n = data / (ch * (bits / 8));
    std::vector<short> raw(n * ch);
    fread(raw.data(), 2, n * ch, f);
    fclose(f);
    out.resize(n);
    for (int i = 0; i < n; i++) {
        long long v = 0;
        for (int c = 0; c < ch; c++) v += raw[i * ch + c];
        v /= ch;
        out[i] = (float)(v / 32768.0);
    }
    return true;
 }
 static bool save_wav(const char* path, const std::vector<float>& x, int sr) {
    FILE* f = fopen(path, "wb");
    if (!f) return false;
    int data = (int)(x.size() * 2);
    char hdr[44];
    memset(hdr, 0, 44);
    memcpy(hdr, "RIFF", 4);
    *(int*)(hdr + 4) = 36 + data;
    memcpy(hdr + 8, "WAVE", 4);
    memcpy(hdr + 12, "fmt ", 4);
    *(int*)(hdr + 16) = 16;
    *(short*)(hdr + 20) = 1;
    *(short*)(hdr + 22) = 1;
    *(int*)(hdr + 24) = sr;
    *(int*)(hdr + 28) = sr * 2;
    *(short*)(hdr + 32) = 2;
    *(short*)(hdr + 34) = 16;
    memcpy(hdr + 36, "data", 4);
    *(int*)(hdr + 40) = data;
    fwrite(hdr, 1, 44, f);
    for (size_t i = 0; i < x.size(); i++) {
        short v = (short)(std::max(-1.0f, std::min(1.0f, x[i])) * 32767.0f);
        fwrite(&v, 2, 1, f);
    }
    fclose(f);
    return true;
 }
 int main(int argc, char** argv) {
    if (argc < 3) { fprintf(stderr, "usage: %s in.wav out.wav [fc,q,sens] ...\n", argv[0]); return 1; }
    std::vector<float> x;
    int sr;
    if (!load_wav(argv[1], x, sr)) { fprintf(stderr, "cannot load %s\n", argv[1]); return 1; }
    std::vector<DetectorBand> bands;
    if (argc >= 4 && strchr(argv[3], ',')) {
        // comma form: one or more "fc,q,sens[,scale]" args, each parsed separately
        for (int i = 3; i < argc; i++) {
            float fc, q, sens, scl = 1.0f;
            if (sscanf(argv[i], "%f,%f,%f,%f", &fc, &q, &sens, &scl) < 3) continue;
            DetectorBand b; b.fc = fc; b.q = q; b.sens = sens; b.level_scale = scl;
            bands.push_back(b);
        }
    } else {
        for (int i = 3; i + 2 < argc; i += 3) {
            DetectorBand b;
            b.fc = (float)atof(argv[i]);
            b.q = (float)atof(argv[i + 1]);
            b.sens = (float)atof(argv[i + 2]);
            if (i + 3 < argc) b.level_scale = (float)atof(argv[i + 3]);
            bands.push_back(b);
        }
    }
    if (bands.empty()) bands.push_back({1000.0f, 1.0f, 12.0f});
    SpectralProcessor sp(2048, 512);
    sp.setDetectorParams(bands);
    std::vector<float> y(x.size());
    const size_t BLK = 1 << 16;
    std::vector<float> inb(BLK), outb(BLK);
    for (size_t s = 0; s < x.size(); s += BLK) {
        size_t n = std::min(BLK, x.size() - s);
        memcpy(inb.data(), x.data() + s, n * sizeof(float));
        for (size_t i = n; i < BLK; i++) inb[i] = 0.0f;
        sp.processBlock(inb.data(), outb.data(), BLK, 1);
        memcpy(y.data() + s, outb.data(), n * sizeof(float));
    }
    save_wav(argv[2], y, sr);
    printf("wrote %s (%zu samples sr=%d, %zu bands)\n", argv[2], y.size(), sr, bands.size());
    return 0;
 }
@@ -1,91 +1,68 @@
 #include <iostream>
 #include <fstream>
 #include <vector>
 #include <string>
 #include <cstring>
 #include <cmath>
 #include <algorithm>
 #include <sstream>
 #include "spectral.hpp"
 #include "filter.hpp"
 #include "detect.hpp"
 #include "ms.hpp"
 #include "params.hpp"
 // WAV16 reader: returns sample rate, fills interleaved float samples (-1..1).
 static float read_wav16(const char* path, std::vector<float>& out) {
    std::ifstream f(path, std::ios::binary);
    if (!f) return -1;
    char riff[12];
    f.read(riff, 12);
-    if (riff[0] != 'R' || riff[1] != 'I' || riff[2] != 'F' || riff[3] != 'F') return -1;
+    if (memcmp(riff, "RIFF", 4) || memcmp(riff + 8, "WAVE", 4)) return -1;
    if (riff[8] != 'W' || riff[9] != 'A' || riff[10] != 'V' || riff[11] != 'E') return -1;
    while (true) {
        char chunk_id[4];
        f.read(chunk_id, 4);
        if (!f.good()) return -1;
        uint32_t chunk_size;
-        f.read(reinterpret_cast<char*>(&chunk_size), 4);
+        if (!f.read(chunk_id, 4) || !f.read(reinterpret_cast<char*>(&chunk_size), 4)) return -1;
-        if (!f.good()) return -1;
+        if (memcmp(chunk_id, "fmt ", 4) == 0) {
        if (chunk_id[0] == 'f' && chunk_id[1] == 'm' && chunk_id[2] == 't' && chunk_id[3] == ' ') {
            if (chunk_size < 16) return -1;
-            
+            uint16_t audio_fmt, channels, block_align, bits;
            int16_t audio_fmt, bits;
            uint16_t channels, block_align;
            uint32_t sample_rate, bytes_per_sec;
            f.read(reinterpret_cast<char*>(&audio_fmt), 2);
            f.read(reinterpret_cast<char*>(&channels), 2);
            f.read(reinterpret_cast<char*>(&sample_rate), 4);
            f.read(reinterpret_cast<char*>(&bytes_per_sec), 4);
            f.read(reinterpret_cast<char*>(&block_align), 2);
            f.read(reinterpret_cast<char*>(&bits), 2);
            if (chunk_size > 16) f.seekg(chunk_size - 16, std::ios::cur);
            while (true) {
-                char data_id[4];
+                char id[4];
-                f.read(data_id, 4);
+                uint32_t dsize;
-                if (!f.good()) return -1;
+                if (!f.read(id, 4) || !f.read(reinterpret_cast<char*>(&dsize), 4)) return -1;
-                
+                if (memcmp(id, "data", 4) == 0) {
-                uint32_t data_size;
+                    size_t n = dsize / (bits / 8);
-                f.read(reinterpret_cast<char*>(&data_size), 4);
+                    out.resize(n);
-                if (!f.good()) return -1;
+                    std::vector<int16_t> raw(n);
-                
+                    f.read(reinterpret_cast<char*>(raw.data()), dsize);
-                if (data_id[0] == 'd' && data_id[1] == 'a' && data_id[2] == 't' && data_id[3] == 'a') {
+                    for (size_t i = 0; i < n; i++) out[i] = static_cast<float>(raw[i]) / 32768.0f;
                    int total = data_size / (bits / 8);
                    out.resize(total);
                    std::vector<int16_t> raw(total);
                    f.read(reinterpret_cast<char*>(raw.data()), data_size);
                    for (int i = 0; i < total; i++) {
                        out[i] = static_cast<float>(raw[i]) / 32768.0f;
                    }
                    return static_cast<float>(sample_rate);
                } else {
-                    f.seekg(data_size, std::ios::cur);
+                    f.seekg(dsize, std::ios::cur);
                }
            }
            break;
        } else {
            f.seekg(chunk_size, std::ios::cur);
        }
    }
    return -1;
 }
-static void write_wav24(const char* path, const float* data, int samples, int channels, int sample_rate) {
+static void write_wav24(const char* path, const float* data, size_t samples, int channels, int sample_rate) {
    std::ofstream f(path, std::ios::binary);
-    
+    int block_align = channels * 3;
-    int bits = 24;
+    int data_size = static_cast<int>(samples) * channels * 3;
    int block_align = channels * bits / 8;
    int bytes_per_sec = sample_rate * block_align;
    int data_size = samples * channels * 3;
    f.write("RIFF", 4);
    int file_size = 36 + data_size;
    f.write("RIFF", 4);
    f.write(reinterpret_cast<const char*>(&file_size), 4);
    f.write("WAVE", 4);
    f.write("fmt ", 4);
    int fmt_size = 16;
    f.write(reinterpret_cast<const char*>(&fmt_size), 4);
@@ -93,45 +70,84 @@ static void write_wav24(const char* path, const float* data, int samples, int ch
    f.write(reinterpret_cast<const char*>(&audio_fmt), 2);
    f.write(reinterpret_cast<const char*>(&channels), 2);
    f.write(reinterpret_cast<const char*>(&sample_rate), 4);
    int bytes_per_sec = sample_rate * block_align;
    f.write(reinterpret_cast<const char*>(&bytes_per_sec), 4);
    f.write(reinterpret_cast<const char*>(&block_align), 2);
    int16_t bits = 24;
    f.write(reinterpret_cast<const char*>(&bits), 2);
    f.write("data", 4);
    f.write(reinterpret_cast<const char*>(&data_size), 4);
-    
+    for (size_t i = 0; i < samples * static_cast<size_t>(channels); i++) {
    for (int i = 0; i < samples * channels; i++) {
        float val = std::max(-1.0f, std::min(1.0f, data[i]));
        int32_t ival = static_cast<int32_t>(val * 8388607.0f);
-        unsigned char bytes[3];
+        unsigned char bytes[3] = { static_cast<unsigned char>(ival & 0xff),
-        bytes[0] = ival & 0xff;
+                                   static_cast<unsigned char>((ival >> 8) & 0xff),
-        bytes[1] = (ival >> 8) & 0xff;
+                                   static_cast<unsigned char>((ival >> 16) & 0xff) };
        bytes[2] = (ival >> 16) & 0xff;
        f.write(reinterpret_cast<const char*>(bytes), 3);
    }
 }
 // Parse `key=value` lines produced by handoff/rpp_allparams.py --flat.
 static PluginParams parse_params_file(const char* path) {
    PluginParams p;
    std::ifstream f(path);
    std::string line;
    BandParams b[6];
    while (std::getline(f, line)) {
        auto eq = line.find('=');
        if (eq == std::string::npos) continue;
        std::string k = line.substr(0, eq);
        double v = std::atof(line.c_str() + eq + 1);
        if (k == "depth") p.depth = v;
        else if (k == "mix") p.mix = v;
        else if (k == "mode") p.mode = v;
        else if (k == "attack") p.attack = v;
        else if (k == "release") p.release = v;
        else if (k == "selectivity") p.selectivity = v;
        else if (k == "sharpness") p.sharpness = v;
        else if (k == "resolution") p.resolution = v;
        else if (k == "offline resolution") p.offline_resolution = v;
        else if (k == "oversample") p.oversample = v;
        else if (k == "offline oversample") p.offline_oversample = v;
        else if (k == "stereo balance") p.stereo_balance = v;
        else if (k == "stereo link") p.stereo_link = v;
        else if (k == "stereo mode") p.stereo_mode = v;
        else if (k == "bypass") p.bypass = v;
        for (int i = 0; i < 6; i++) {
            std::string pre = "band" + std::to_string(i) + " ";
            if (k == pre + "freq") b[i].freq = v;
            else if (k == pre + "q") b[i].q = v;
            else if (k == pre + "sens") b[i].sens = v;
            else if (k == pre + "mode") b[i].mode = v;
            else if (k == pre + "on") b[i].on = v;
            else if (k == pre + "balance") b[i].balance = v;
        }
    }
    for (auto& bd : b) p.bands.push_back(bd);
    return p;
 }
 int main(int argc, char* argv[]) {
    if (argc < 3) {
-        std::cerr << "Usage: " << argv[0] << " input.wav output.wav" << std::endl;
+        std::cerr << "Usage: " << argv[0] << " input.wav output.wav [params.conf]\n";
        return 1;
    }
    PluginParams params;
    if (argc > 3) params = parse_params_file(argv[3]);
    else {
        params.bands.push_back(BandParams{});
    }
    std::vector<float> input;
    float sr = read_wav16(argv[1], input);
    if (sr <= 0 || input.empty()) {
-        std::cerr << "Failed to read input file" << std::endl;
+        std::cerr << "Failed to read input file\n";
        return 1;
    }
-    
+    int channels = 2;  // all etalon renders are 2ch
-    size_t total_samples = input.size();
+    size_t frames = input.size() / static_cast<size_t>(channels);
-    int channels = 2;
+    // Trim guard: output length == input length (honest metric, B.14).
-    size_t frames = total_samples / channels;
+    if (input.size() % channels != 0) frames = input.size() / channels;
    std::vector<float> output(total_samples);
    SpectralProcessor sp(2048, 512);
    sp.setDetectorParams(10.0f, 10.0f, 0.864f);
    std::vector<float> left_in(frames), right_in(frames);
    for (size_t i = 0; i < frames; i++) {
@@ -139,20 +155,35 @@ int main(int argc, char* argv[]) {
        right_in[i] = input[i * 2 + 1];
    }
    SpectralProcessor sp(2048, 512);
    std::vector<DetectorBand> bands;
    for (const auto& b : params.bands) {
        if (b.on > 0.5f && b.freq > 1.0f) {
            DetectorBand db;
            db.fc = static_cast<float>(b.freq);
            db.q = static_cast<float>(b.q > 0.0f ? b.q : 1.0f);
            db.sens = static_cast<float>(b.sens);
            bands.push_back(db);
        }
    }
    if (bands.empty()) {
        DetectorBand db{500.0f, 1.0f, 12.0f};
        bands.push_back(db);
    }
    sp.setDetectorParams(bands);
    std::vector<float> left(frames, 0.0f), right(frames, 0.0f);
    encode_ms(left_in.data(), right_in.data(), frames);
    sp.processBlock(left_in.data(), left.data(), frames, 1);
    sp.processBlock(right_in.data(), right.data(), frames, 1);
    decode_ms(left.data(), right.data(), frames);
    std::vector<float> output(frames * 2);
    for (size_t i = 0; i < frames; i++) {
        output[i * 2] = left[i];
        output[i * 2 + 1] = right[i];
    }
    write_wav24(argv[2], output.data(), frames, channels, static_cast<int>(sr));
-    
+    std::cout << "Done! frames=" << frames << " sr=" << sr << "\n";
    std::cout << "Done!" << std::endl;
    return 0;
 }
@@ -24,7 +24,7 @@
 #include <cstdint>
 // Constants extracted from binary
-static constexpr float SCALE = 0.0009775171056389809f;  // 1/1024 (DAT_1824c3c54)
+static constexpr float SCALE = 0.0009775171056389809f;  // 1/1023 (DAT_1824c3c54, verified 2026-08-19)
 static constexpr float ONE   = 1.0f;                     // DAT_1824c3ea4
 static constexpr float TWO   = 2.0f;                     // DAT_1824c41e0
 static constexpr float NEG1  = -1.0f;                    // DAT_1824c4680
@@ -33,7 +33,10 @@ static constexpr float ZERO  = 0.0f;                     // DAT_1824c4140
 static constexpr float DEPTH_SCALE = 4.0f;               // DAT_1824c4334
 static constexpr float DB_CONV = 8.68588924407959f;      // 20/ln(10) (DAT_1824c43e0)
 static constexpr float FLOOR_DB = -6.907755374908447f;   // ln(0.001) (DAT_1824c4704)
-static constexpr float FLOOR LIN = 0.001f;               // exp(FLOOR_DB)
+static constexpr float FLOOR_LIN = 0.001f;               // exp(FLOOR_DB)
 static constexpr double TWO_PI = 6.283185307179586;      // DAT_1824c4248 (2π, twin-mask factory)
 static constexpr float SCALE_1024 = 0.0009765625f;       // 1/1024 (DAT_1824c3c50, band LUT apply)
 static constexpr float CONST_5 = 5.0f;                   // DAT_1824c4230 (AudioProcessingModule ctor)
 // PRNG state offsets from param_1
 static constexpr int PRNG_STATE = 0x2404e0;
@@ -51,108 +54,125 @@ struct BandConfig {
    void* callback;  // +0x50: vtable callback (if non-null, use callback)
 };
-// LUT evaluation for a single bin
+// Structural LUT curve (f_563440.dis, exact transcription 2026-08-19)
-// x is in [0, 1] range
+// x in [0,1], gamma == band->threshold (offset +0x0c), A=+0x00, B=+0x04
 static float eval_lut_bin(float x, const BandConfig* band) {
-    // Path 1: callback exists → use vtable
+    float gamma = band->threshold;
-    if (band->callback != nullptr) {
+    float result;
-        // TODO: transcribe callback vtable call
+    if (band->flag == 0) {
-        return x;
+        // Linear path (0x563595): t = x^(1/γ) if γ!=1 && x>0; val = A + (B-A)*t
-    }
+        // decomp: fVar16 = expf(logf(x)/gamma)  (FLOAT log/exp)
-    
+        float t = x;
-    // Path 2: power-law (flag != 0 and threshold != 1.0)
+        if (gamma != ONE && x > ZERO) {
-    if (band->flag != 0 && band->threshold != ONE) {
+            t = expf(logf(x) / gamma);
        float C = band->threshold;
        // x = 2*x - 1 (center at zero: [-1, 1])
        float centered = TWO * x - ONE;
        if (C == ONE || centered == ZERO) {
            // fall through to linear
        } else {
            // sign(x) * 10^(log10(|x|) / C)
            float sign = (centered < ZERO) ? NEG1 : ONE;
            // absolute value: |x|
            float abs_x = fabsf(centered);
            // if abs_x > 0: result = sign * exp(log(|x|) * (1/C))
            if (abs_x > ZERO) {
                float log_val = log10f(abs_x);
                float result = powf(10.0f, log_val / C);
                centered = sign * result;
            }
            // fall through to linear with transformed x
            x = centered * HALF + HALF;  // remap back to [0,1]
        }
        result = band->A + (band->B - band->A) * t;
    } else {
        // Power-law path (0x5635cd): t = 2x-1; if γ!=1 && t!=0: t = sign(t)·|t|^(1/γ)
        // val = A + (B-A)·0.5·(1+t); decomp: sign·expf(logf(|t|)/gamma)
        float t = TWO * x - ONE;
        if (gamma != ONE && t != ZERO) {
            float sign = (t < ZERO) ? NEG1 : ONE;
            t = expf(logf(fabsf(t)) / gamma) * sign;
        }
        result = band->A + (band->B - band->A) * HALF * (ONE + t);
    }
-    
+    return result;
    // Path 3: linear interpolation (always applied after transform)
    float slope = band->B - band->A;
    return slope * x + band->A;
 }
-// FUN_180563440: LUT curve evaluation for 1024 bins
+// FUN_180563440: LUT curve evaluation for 0x400 bins
-// r13 = context pointer (param_1)
+// r13 = context pointer (param_1). Loop counter edi, x = i*SCALE clamp[0,1],
-// Reads: band config at r13+0x188 (one per band)
+// band config read from r13+0x188 each iteration (rbx), output double at r13+0x198[i*8].
-// Writes: output at r13+0x198 (1024 doubles, stride 8)
+void lut_curve_eval(void* ctx) {
 void lut_curve_eval(void* ctx, int bin_start, int bin_end) {
    auto* base = static_cast<uint8_t*>(ctx);
    int band_count = *reinterpret_cast<int*>(base + 0x540868);
    if (band_count <= 0) {
        // Initialize with default 0x800 bins
        band_count = 0x800;  // 2048? or 1024?
    }
    // Output pointer: r13+0x198
    double* output = reinterpret_cast<double*>(base + 0x198);
-    
+    BandConfig* band = reinterpret_cast<BandConfig*>(base + 0x188);
-    // Evaluate LUT curve for each bin (0x400 = 1024 iterations)
+    for (int bin = 0; bin < 0x400; bin++) {   // cmp $0x400 jl
    for (int bin = 0; bin < 0x400; bin++) {
        float x = static_cast<float>(bin) * SCALE;
-        x = fminf(fmaxf(x, ZERO), ONE);  // clamp to [0, 1]
+        x = fminf(x, ONE);
-        
+        if (x < ZERO) x = ZERO;
-        BandConfig* band = reinterpret_cast<BandConfig*>(base + 0x188);
+        output[bin] = static_cast<double>(eval_lut_bin(x, band));
        float result = eval_lut_bin(x, band);
        // Store as double-precision (line 196: cvtss2sd + movsd [rsi])
        output[bin] = static_cast<double>(result);
    }
 }
 // FUN_18056e3e0: twin-mask factory
-// Creates per-band mask by applying twin resonance to the LUT curve
+// DECODED (decomp_funs2.txt:7988 + f_56e3e0.dis): fills the 0x400-bin mask
-// band_count = number of bands (max 6)
+// with a SINGLE scalar  s = 2π / (count·SR), where
-// N = 1024 (FFT size for LUT evaluation)
+//   count = [ctx+0x240080] (int),  SR = [ctx+0x24] (float, internal SR).
-// Output stride: 0x2000 (8192 bytes = 1024 doubles)
+// NOT a per-bin twin resonance — a constant fill (the "twin" shape enters
 // elsewhere via the LUT curve FUN_180563440). Output mask stride 0x2000/band.
 void twin_mask_factory(void* ctx, int band_idx, int n_bins) {
    auto* base = static_cast<uint8_t*>(ctx);
-    // Calls twin evaluation for each bin
+    int count = *reinterpret_cast<int*>(base + 0x240080);
-    // TODO: transcribe the full loop from disassembly
+    float sr = *reinterpret_cast<float*>(base + 0x24);
-    // The factory applies the band's resonance shape to the LUT curve
+    double s = TWO_PI / (static_cast<double>(count) * static_cast<double>(sr));
    float* mask = reinterpret_cast<float*>(base + 0x4198 + band_idx * 0x2000);
    for (int i = 0; i < 0x400; i++) {
        mask[i] = static_cast<float>(s);
    }
 }
-// FUN_180563a60: band combine
+// FUN_180563a60: band LUT apply (level -> gain). DECODED (decomp_funs2.txt:8975 + f_563a60.dis).
-// Combines 6 band masks into final per-bin gain
+// For each of 6 bands and 0x400 bins:
-// Stereo: max 2 channels, output stride per band = 0x2000
+//   level_dB   = 20·log10(mask[band][bin])          (logf · 8.6859)
-// Pattern: gain = 1.0 - sum(band_masks)
+//   level_axis[bin] = bin·(1/1024)                   (SCALE_1024)
-void band_combine(void* ctx, int n_channels, int n_bins) {
+//   t          = clamp((dB − A)/(B − A), 0, 1)        (BandConfig ctx+0x180: A,B,gamma,flag)
 //   if gamma == 1.0:          val = t
 //   elif flag == 0 (linear):  val = t^gamma            (powf, NOT 1/gamma)
 //   else (power-law):          val = 0.5·(1 + sign(2t−1)·|2t−1|^gamma)
 //   level_axis[bin+1] = val   (pairs level, gain)
 // NOTE: this is the INVERSE curve of FUN_180563440 (which uses x^(1/γ)).
 void band_lut_apply(void* ctx) {
    auto* base = static_cast<uint8_t*>(ctx);
-    int band_count = *reinterpret_cast<int*>(base + 0x540868);
+    float* bandcfg = *reinterpret_cast<float**>(base + 0x180);
-    if (band_count > 6) band_count = 6;
+    float A = bandcfg[0];
-    
+    float B = bandcfg[1];
-    // Output accumulator at r13+0x2198
+    float gamma = bandcfg[3];
-    // Each band's mask is at r13+0x2198 + band_idx * 0x2000
+    float flag = bandcfg[4];
-    
+    double* mask = reinterpret_cast<double*>(base + 0x4198);
-    for (int ch = 0; ch < n_channels; ch++) {
+    for (int band = 0; band < 6; band++) {
-        // For each bin: sum all band contributions
+        double* m = mask + band * 0x400;
-        // Then invert: gain = 1.0 - sum
+        float* level_gain = *reinterpret_cast<float**>(base + 0xe0 + band * 0x18);
-        double* acc = reinterpret_cast<double*>(base + 0x2198 + ch * 0x2000);
+        for (int bin = 0; bin < 0x400; bin++) {
-        for (int bin = 0; bin < n_bins; bin++) {
+            float db = logf(static_cast<float>(m[bin])) * DB_CONV;
-            acc[bin] = ONE - acc[bin];
+            level_gain[bin * 2] = static_cast<float>(bin) * SCALE_1024;
            float t = (db - A) / (B - A);
            t = std::max(ZERO, std::min(ONE, t));
            float val = t;
            if (gamma != ONE) {
                if (flag == ZERO) {
                    val = powf(t, gamma);
                } else {
                    float u = TWO * t - ONE;
                    float sgn = (u < ZERO) ? NEG1 : ONE;
                    val = HALF * (ONE + sgn * powf(fabsf(u), gamma));
                }
            }
            level_gain[bin * 2 + 1] = val;
        }
    }
 }
 // ---- mask-accumulator combine kernels (FUN_180529fe0, CRT thunks) ----
 // Signatures recovered from raw bytes in the rt snap (objdump of 0x180008d60/5a20/3c40).
 //
 // 0x8d60 combine3: out[i] = a[i] - b[i]   (vsubpd, 3 pointers; dst is the 3rd arg)
 //   In the per-band loop: 0x5406f8[i] = 0x540678[i] - 0x5407c8[i]
 void combine_sub(double* out, const double* a, const double* b, int n) {
    for (int i = 0; i < n; i++) out[i] = a[i] - b[i];
 }
 // 0x5a20: dst[i] += src[i]  (double; kernel 0x18001a5a0)
 void acc_add(double* dst, const double* src, int n) {
    for (int i = 0; i < n; i++) dst[i] += src[i];
 }
 // 0x3c40: dst[i] += a[i] * b[i]  (double; vfmadd213pd)
 void acc_fma(double* dst, const double* a, const double* b, int n) {
    for (int i = 0; i < n; i++) dst[i] += a[i] * b[i];
 }
 // FUN_180529fe0: coefficient setup (from decomp_funs.txt)
 // Generates per-band coefficients via PRNG, applies depth scaling
 // This is the vtable method for Soothe2Module
 void coefficient_setup(void* ctx, int band_idx, int param3, int param4) {
    auto* base = static_cast<uint8_t*>(ctx);
@@ -0,0 +1,18 @@
 #pragma once
 // Level-path transcriptions (decomp_funs2.txt + f_*.dis).
 // All offsets are relative to the level-path object (NOT the DSP ctx).
 // FUN_180563440: LUT curve build (x = i/1023, linear x^(1/gamma), power-law).
 void lut_curve_eval(void* ctx);
 // FUN_18056e3e0: twin-mask factory = constant fill 2π/(count·SR) over 0x400 bins.
 void twin_mask_factory(void* ctx, int band_idx, int n_bins);
 // FUN_180563a60: band LUT apply (level -> gain) t^gamma / power-law, 6 bands.
 void band_lut_apply(void* ctx);
 // mask-accumulator combine kernels (FUN_180529fe0 CRT thunks).
 void combine_sub(double* out, const double* a, const double* b, int n);  // 0x8d60
 void acc_add(double* dst, const double* src, int n);                      // 0x5a20
 void acc_fma(double* dst, const double* a, const double* b, int n);       // 0x3c40
@@ -0,0 +1,68 @@
 #include <cstdio>
 #include <cmath>
 #include <cstring>
 #include <cstdint>
 #include <vector>
 #include "levelpath.hpp"
 int main() {
    bool ok = true;
    auto chk = [&](const char* n, double got, double exp, double tol) {
        bool p = std::fabs(got - exp) < tol;
        ok = ok && p;
        std::printf("  %-28s got=%.6f exp=%.6f (%s)\n", n, got, exp, p ? "ok" : "MISMATCH");
    };
    // ---- twin_mask_factory (FUN_18056e3e0): fill 2π/(count·SR) ----
    {
        std::vector<uint8_t> b(0x241000, 0);
        *reinterpret_cast<int*>(b.data() + 0x240080) = 1024;
        *reinterpret_cast<float*>(b.data() + 0x24) = 48000.0f;
        twin_mask_factory(b.data(), 0, 1024);
        float expect = static_cast<float>(6.283185307179586 / (1024.0 * 48000.0));
        float* mask = reinterpret_cast<float*>(b.data() + 0x4198);
        chk("twin_mask_factory fill", mask[0], expect, 1e-9);
        chk("twin_mask_factory uniform", mask[511], expect, 1e-9);
    }
    // ---- band_lut_apply (FUN_180563a60): t^gamma linear curve ----
    {
        std::vector<uint8_t> b(0x12000, 0);
        float bandcfg[8] = {0.0f, 1.0f, 0.0f, 2.0f, 0.0f, 0, 0, 0};  // A=0,B=1,gamma=2,flag=0
        *reinterpret_cast<float**>(b.data() + 0x180) = bandcfg;
        std::vector<float> level_gain(0x800 * 6, 0.0f);
        for (int band = 0; band < 6; band++) {
            *reinterpret_cast<float**>(b.data() + 0xe0 + band * 0x18) = level_gain.data() + band * 0x800;
        }
        double* mask = reinterpret_cast<double*>(b.data() + 0x4198);
        // mask[0] = 1.0  -> dB=0 -> t=(0-0)/(1-0)=0 -> val=0^2=0
        // mask[1] = 0.3162277 (=-10dB) -> t=(-10-0)/1=-10 -> clamp 0 -> 0
        // mask[2] = 1.0  -> 0
        for (int i = 0; i < 0x400 * 6; i++) mask[i] = 1.0;
        // one bin at dB = +0.5 (mask = 10^(0.5/20) = 1.059254) -> t = 0.5 -> 0.5^2 = 0.25
        mask[0] = 1.0592537251772883;  // +0.5 dB
        band_lut_apply(b.data());
        chk("band_lut_apply level_axis[0]", level_gain[0], 0.0f, 1e-6);   // bin 0 * 1/1024
        chk("band_lut_apply gain[0] t^gamma", level_gain[1], 0.25f, 1e-3); // 0.5^2
        // bin at index 512: level = 512/1024 = 0.5
        chk("band_lut_apply level_axis[512]", level_gain[1024], 0.5f, 1e-6);
    }
    // ---- combine kernels (0x8d60/0x5a20/0x3c40) ----
    {
        double out[4] = {0, 0, 0, 0};
        double a[4] = {5, 2, -1, 8};
        double b[4] = {3, 7, 4, 2};
        combine_sub(out, a, b, 4);
        chk("combine_sub[0] a-b", out[0], 2.0, 1e-12);
        chk("combine_sub[2] a-b", out[2], -5.0, 1e-12);
        acc_add(out, a, 4);  // out = (a-b) + a
        chk("acc_add[0]", out[0], 7.0, 1e-12);   // (5-3)+5 = 7
        acc_fma(out, b, a, 4);  // out += b*a
        // out[1] = (2-7)+2 = -3, then -3 + b[1]*a[1] = -3 + 7*2 = 11
        chk("acc_fma[1]", out[1], 11.0, 1e-12);
    }
    std::printf(ok ? "ALL OK\n" : "FAILURES\n");
    return ok ? 0 : 1;
 }
@@ -0,0 +1,42 @@
 #include "leveltrack.hpp"
 namespace leveltrack {
 void iir_first_order(float* x, const double* A, size_t n) {
    double acc = 0.0;
    for (size_t i = 0; i < n; i++) {
        double y = static_cast<double>(x[i]) * A[i] + acc;
        acc = y;
        x[i] = static_cast<float>(y);
    }
 }
 void iir_bidirectional(float* x, const double* A, size_t n) {
    iir_first_order(x, A, n);
    // reverse pass over reversed indices back into x (keep array order).
    double acc = 0.0;
    for (size_t k = n; k-- > 0;) {
        double y = static_cast<double>(x[k]) * A[k] + acc;
        acc = y;
        x[k] = static_cast<float>(y);
    }
 }
 void iir_first_order_unrolled4(float* x, const double* A, size_t n) {
    double acc = 0.0;
    size_t i = 0;
    for (; i + 4 <= n;) {
        for (int u = 0; u < 4; u++, i++) {
            double y = static_cast<double>(x[i]) * A[i] + acc;
            acc = y;
            x[i] = static_cast<float>(y);
        }
    }
    for (; i < n; i++) {
        double y = static_cast<double>(x[i]) * A[i] + acc;
        acc = y;
        x[i] = static_cast<float>(y);
    }
 }
 }  // namespace leveltrack
@@ -0,0 +1,23 @@
 #pragma once
 #include <cstddef>
 // Level-tracker UPDATE (structural transcription, NOTES_LEVEL 2026-08-19f).
 // The plugin smooths the per-bin level mask with first-order "leaky-integrator"
 // IIR stages (cumulative running-acc) in FUN_180529fe0. Three stages run in the
 // real per-band loop; the coefficient arrays A[] are runtime values (from the
 // level sidechain / smoothing-param builder). This module provides the exact
 // scalar and vectorized forms as pure functions; callers supply A[].
 namespace leveltrack {
 // Scalar first-order stage:  acc=0; y[i]=x[i]*A[i]+acc; acc=y; x[i]=y
 // Mirrors consumers_out.txt scalar loop. In-place on x.
 void iir_first_order(float* x, const double* A, size_t n);
 // Bidirectional stage: forward then reverse pass over x with same A.
 void iir_bidirectional(float* x, const double* A, size_t n);
 // Vectorized-wide variant handling 4 elements per iteration (like the plugin's
 // unrolled loop). Same math, provided for parity tests.
 void iir_first_order_unrolled4(float* x, const double* A, size_t n);
 }  // namespace leveltrack
@@ -0,0 +1,40 @@
 #include <cstdio>
 #include <cmath>
 #include <vector>
 #include <cstring>
 #include "leveltrack.hpp"
 #include "leveltrack_data.hpp"
 int main() {
    const size_t n = 37;  // odd size to exercise the unrolled tail.
    std::vector<double> A(n);
    std::vector<float> x1(n), x2(n);
    for (size_t i = 0; i < n; i++) {
        A[i] = 0.05 + 0.9 * double(1 + i % 3) / 3.0;  // nontrivial profile
        x1[i] = static_cast<float>(0.5 * std::sin(0.3 * i) * (1 + i * 0.01));
    }
    std::memcpy(x2.data(), x1.data(), n * sizeof(float));
    leveltrack::iir_first_order(x1.data(), A.data(), n);
    leveltrack::iir_first_order_unrolled4(x2.data(), A.data(), n);
    double maxd = 0.0;
    for (size_t i = 0; i < n; i++) maxd = std::fmax(maxd, std::fabs(x1[i] - x2[i]));
    std::printf("scalar vs unrolled4 max|dx| = %.3e  (%s)\n",
                maxd, maxd < 1e-6 ? "OK" : "MISMATCH");
    // bidirectional: equals applying forward then reverse.
    std::vector<float> rx(n);
    leveltrack::iir_bidirectional(rx.data(), A.data(), n);
    std::printf("leveltrack integration check done (n=%zu)\n", n);
    // P1.5 live-capture data sanity (render_long.rpp: attack=0 release=0).
    std::printf("live ltk: SR=%.0f nbins=%zu  A_ATTACK[1]=%.4f A_ATTACK[319]=%.4f "
                "A_RELEASE[1]=%.6f scalar870=%.2f scalar884=%.2f scalar888=%.2f\n",
                (double)ltk::CTX_INTERNAL_SR, ltk::LEVEL_NBINS,
                ltk::A_ATTACK[1], ltk::A_ATTACK[319],
                ltk::A_RELEASE[1], (double)ltk::SCALAR0X40870,
                (double)ltk::SCALAR0X40884, (double)ltk::SCALAR0X40888);
    return maxd < 1e-6 ? 0 : 1;
 }
@@ -0,0 +1,149 @@
 // AUTOGENERATED from P1.5 realtime capture handoff/rtctx_live.json
 // (render_long.rpp: attack=0 release=0 selectivity=10 sharpness=10 depth=0.864).
 // Regenerate with handoff/emit_leveltrack.py. Do not edit by hand.
 #pragma once
 #include <cstddef>
 namespace ltk {
 // ctx = 0x2370040, marker +0x24 == 48000.0f (internal SR).
 constexpr float CTX_INTERNAL_SR = 48000.0f;
 // mask scalars (float) captured live:
 constexpr float SCALAR0X40870 = 440.9548645019531f;
 constexpr float SCALAR0X40874 = 1.0f;
 constexpr float SCALAR0X40878 = 1.0f;
 constexpr float SCALAR0X4087C = 1.0f;
 constexpr float SCALAR0X40880 = 25.000001907348633f;
 constexpr float SCALAR0X40884 = 10.0f;
 constexpr float SCALAR0X40888 = 1.0f;
 constexpr float SCALAR0X4088C = 1.0f;
 constexpr float SCALAR0X40890 = 0.0f;
 constexpr float SCALAR0X40894 = 1200.0001220703125f;
 constexpr float SCALAR0X408AC = 9.183549615799121e-41f;
 constexpr size_t LEVEL_NBINS = 341;
 // per-bin level-tracker IIR attack coefficients (0x4c0528):
 const double A_ATTACK[341] = {
  0, 0.34005138278007507, 0.3483593761920929, 0.35649341344833374, 0.3644568920135498, 0.37225314974784851,
  0.37988573312759399, 0.38735818862915039, 0.39467430114746094, 0.40183743834495544, 0.40885132551193237, 0.41571962833404541,
  0.4224458634853363, 0.42903351783752441, 0.43548604846000671, 0.44180700182914734, 0.44799956679344177, 0.45406708121299744,
  0.46001288294792175, 0.46583998203277588, 0.47155144810676575, 0.47715029120445251, 0.48263949155807495, 0.48802188038825989,
  0.49330011010169983, 0.49847698211669922, 0.50355511903762817, 0.50853699445724487, 0.51342517137527466, 0.51822203397750854,
  0.52292978763580322, 0.52755081653594971, 0.53208738565444946, 0.53654158115386963, 0.54091531038284302, 0.54521089792251587,
  0.54942995309829712, 0.55357468128204346, 0.55764675140380859, 0.56164795160293579, 0.56558007001876831, 0.56944471597671509,
  0.57324361801147461, 0.57697826623916626, 0.58065032958984375, 0.58426117897033691, 0.58781230449676514, 0.59130507707595825,
  0.59474086761474609, 0.59812110662460327, 0.60144698619842529, 0.60471975803375244, 0.60794061422348022, 0.61111080646514893,
  0.6142314076423645, 0.61730360984802246, 0.62032842636108398, 0.6233069896697998, 0.62624025344848633, 0.62912911176681519,
  0.63197475671768188, 0.63477796316146851, 0.63753974437713623, 0.64026087522506714, 0.64294230937957764, 0.64558488130569458,
  0.64818942546844482, 0.65075665712356567, 0.65328741073608398, 0.65578234195709229, 0.65824240446090698, 0.66066807508468628,
  0.66306018829345703, 0.66541939973831177, 0.66774630546569824, 0.67004162073135376, 0.67230594158172607, 0.67453992366790771,
  0.67674410343170166, 0.67891907691955566, 0.68106538057327271, 0.68318367004394531, 0.68527436256408691, 0.68733805418014526,
  0.68937522172927856, 0.69138640165328979, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
  0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274, 0.69213396310806274,
 };
 // per-bin level-tracker IIR release coefficients (0x3c0510 == 0x2c04f8):
 const double A_RELEASE[341] = {
  0, 0.00075329304672777653, 0.00088481028797104955, 0.0010319935390725732, 0.0011957522947341204, 0.0013769635697826743,
  0.0015764719573780894, 0.0017950870096683502, 0.0020335691515356302, 0.0022926407400518656, 0.002572979312390089, 0.002875214209780097,
  0.0031999291386455297, 0.003547657048329711, 0.0039188801310956478, 0.0043140365742146969, 0.0047335159033536911, 0.005177660845220089,
  0.0056467670947313309, 0.0061410837806761265, 0.0066608106717467308, 0.0072061149403452873, 0.0077771246433258057, 0.0083739152178168297,
  0.0089965220540761948, 0.0096449656412005424, 0.010319218970835209, 0.011019209399819374, 0.01174485869705677, 0.012496041133999825,
  0.013272601179778576, 0.01407436840236187, 0.014901150017976761, 0.015752725303173065, 0.016628840938210487, 0.017529245465993881,
  0.018453648313879967, 0.019401764497160912, 0.020373275503516197, 0.021367868408560753, 0.022385178133845329, 0.023424861952662468,
  0.024486592039465904, 0.025569943711161613, 0.026674600318074226, 0.027800126001238823, 0.028946168720722198, 0.030112311244010925,
  0.031298160552978516, 0.032503306865692139, 0.033727359026670456, 0.034969881176948547, 0.036230511963367462, 0.037508808076381683,
  0.038804333657026291, 0.040116727352142334, 0.041445545852184296, 0.042790420353412628, 0.044150922447443008, 0.045526620000600815,
  0.046917147934436798, 0.048322096467018127, 0.049741055816411972, 0.051173664629459381, 0.052619524300098419, 0.054078217595815659,
  0.055549412965774536, 0.057032708078622818, 0.05852774903178215, 0.060034122318029404, 0.061551533639431, 0.063079580664634705,
  0.064617909491062164, 0.066166229546070099, 0.067724093794822693, 0.069291271269321442, 0.070867374539375305, 0.072452105581760406,
  0.074045136570930481, 0.07564612478017807, 0.077254779636859894, 0.078870825469493866, 0.080493956804275513, 0.082123853266239166,
  0.08376021683216095, 0.085402816534042358, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
  0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808, 0.086020313203334808,
 };
 }  // namespace ltk
@@ -0,0 +1,48 @@
 #include "log2_ln.hpp"
 #include <cstring>
 #include <cmath>
 namespace soothe2 {
 float ln_plugin_f32(float x) {
    if (x <= 0.0f) return -INFINITY;
    uint32_t bits;
    std::memcpy(&bits, &x, sizeof(uint32_t));
    int exp = int((bits >> 23) & 0xFF);
    uint32_t mantissa = bits & 0x7FFFFFu;
    if (exp == 0) return -INFINITY;
    float x_norm = mantissa * (1.0f / 8388608.0f);
    constexpr float c0_a = -0.1517720520f;
    constexpr float c0_b =  0.1696488112f;
    constexpr float c1   = -0.1646245718f;
    constexpr float c2   =  0.1982250363f;
    constexpr float c3   = -0.2500466406f;
    constexpr float c4   =  0.3333656490f;
    constexpr float c5   = -0.5000000000f;
    constexpr float ln2  =  0.6931471825f;
    constexpr float c0_init = c0_a * c0_b;
    float y = c0_init + x_norm;
    y = y * x_norm + c1;
    y = y * x_norm + c2;
    y = y * x_norm + c3;
    y = y * x_norm + c4;
    y = y * x_norm + c5;
    float ln_m = x_norm + x_norm * x_norm * y;
    return ln2 * float(exp - 127) + ln_m;
 }
 void ln_plugin_f32_arr(const float* in, float* out, size_t n) {
    for (size_t i = 0; i < n; ++i) {
        out[i] = ln_plugin_f32(in[i]);
    }
 }
 }  // namespace soothe2
@@ -0,0 +1,17 @@
 #pragma once
 #include <cstdint>
 #include <cmath>
 #include <cstring>
 namespace soothe2 {
 // Plugin's exact ln(float) from 0x1802a24c0 (535a70 FFT-conv engine)
 // Computes natural logarithm via mantissa polynomial + exponent scaling
 // Coefficients extracted from binary at 0x181f81f80..0x181f821c0
 // Max error ~3e-6 for typical inputs (x in [1, 1.34))
 float ln_plugin_f32(float x);
 // Vectorized version for arrays
 void ln_plugin_f32_arr(const float* in, float* out, size_t n);
 }  // namespace soothe2
@@ -0,0 +1,38 @@
 #pragma once
 // Plugin parameters decoded from RPP <SOOTHE2STATE> XML (see handoff/rpp_allparams.py).
 #include <string>
 #include <vector>
 #include <cmath>
 struct BandParams {
    double balance = 0.5;
    double freq = 678.7611083984375;
    double mode = 1.0;
    double on = 0.0;
    double q = 0.9999978542327881;
    double sens = 12.0;
 };
 struct PluginParams {
    double attack = 0.0;
    double delta = 0.0;
    double depth = 0.8639736175537109;
    double mix = 100.0;
    double mode = 1.0;
    double oversample = 0.0;
    double offline_oversample = 3.0;
    double offline_resolution = 4.0;
    double release = 0.0;
    double resolution = 1.0;
    double selectivity = 10.0;
    double sharpness = 10.0;
    double bypass = 0.0;
    double input_trim = 0.0;
    double trim = 0.0;
    double sidechain = 0.0;
    double sidechain_solo_on = 0.0;
    double stereo_balance = 0.2840004563331604;
    double stereo_link = 100.0;
    double stereo_mode = 1.0;
    std::vector<BandParams> bands;  // up to 6
 };
@@ -0,0 +1,238 @@
 // render48k.cpp — 48000/N=4096 internal-grid stereo renderer
 //
 // Host audio is 44100; the plugin detector runs internally at 48000/N=4096 (the
 // live IIR/warp/freq-axis tables are sized for that grid). This tool mirrors that:
 //   1. read input WAV (44100 host samples, stereo or mono)
 //   2. resample 44100 -> 48000  (libsamplerate, SINC best)
 //   3. Process L and R channels (stereo link=100%: same processing for both)
 //   4. Apply balance: scale reduction for R channel
 //   5. Apply mix: wet-dry mix
 //   6. resample 48000 -> 44100
 //   7. write 24-bit output WAV (stereo)
 //
 // Usage:
 //   render48k <in.wav> <out.wav> fc,q,sens[,scale] ...
 // Env:
 //   RT_STEREO_LINK=1.0   (1.0 = sum channels for analysis, 0.0 = dual mono)
 //   RT_STEREO_BALANCE=0.284  (R channel reduction scale, 1.0 = equal, <1.0 = less on R)
 //   RT_DEPTH=0.864       (sens multiplier)
 //   RT_MIX=1.0           (0=dry, 1=full wet)
 #include "spectral.hpp"
 #include <cstdio>
 #include <cstdlib>
 #include <vector>
 #include <cmath>
 #include <cstring>
 #include <samplerate.h>
 static int g_in_ch = 1;
 static bool load_wav_stereo(const char* path, std::vector<float>& out, int& sr, int& channels) {
    FILE* f = fopen(path, "rb");
    if (!f) return false;
    char hdr[44];
    if (fread(hdr, 1, 44, f) != 44) return false;
    sr = *(int*)(hdr + 24);
    int ch = *(short*)(hdr + 22);
    int bits = *(short*)(hdr + 34);
    int data = 0;
    int64_t pos = 12;
    fseek(f, 12, SEEK_SET);
    while (pos < 32 * 1024 * 1024) {
        char cid[4]; int csize;
        if (fread(cid, 1, 4, f) < 4 || fread(&csize, 4, 1, f) < 1) break;
        pos += 8;
        if (memcmp(cid, "data", 4) == 0) { data = csize; break; }
        pos += csize;
        int skip = csize;
        if (csize % 2) skip++;
        fseek(f, skip, SEEK_CUR);
    }
    if (!data) { fclose(f); return false; }
    int n = data / (ch * (bits / 8));
    g_in_ch = ch;
    channels = 2;
    out.resize(n * 2);
    if (bits == 16) {
        std::vector<short> raw(n * ch);
        fread(raw.data(), 2, n * ch, f);
        for (int i = 0; i < n; i++) {
            float v = 0.0f;
            if (ch == 1) {
                v = raw[i] / 32768.0f;
                out[2*i] = v; out[2*i+1] = v;
            } else {
                out[2*i] = raw[2*i] / 32768.0f;
                out[2*i+1] = raw[2*i+1] / 32768.0f;
            }
        }
    } else if (bits == 24) {
        std::vector<unsigned char> raw(n * ch * 3);
        fread(raw.data(), 1, n * ch * 3, f);
        for (int i = 0; i < n; i++) {
            auto read24 = [&](int idx) -> float {
                int32_t s = (raw[idx] | (raw[idx+1] << 8) | (raw[idx+2] << 16));
                if (s & 0x800000) s |= 0xFF000000;
                return s / 8388608.0f;
            };
            if (ch == 1) {
                float v = read24(i * 3);
                out[2*i] = v; out[2*i+1] = v;
            } else {
                out[2*i] = read24(2*i * 3);
                out[2*i+1] = read24((2*i+1) * 3);
            }
        }
    } else return false;
    fclose(f);
    return true;
 }
 static bool save_wav24_stereo(const char* path, const std::vector<float>& L, const std::vector<float>& R, int sr) {
    FILE* f = fopen(path, "wb");
    if (!f) return false;
    int ch = 2, bits = 24;
    size_t n = std::min(L.size(), R.size());
    int data = (int)(n * ch * (bits / 8));
    char hdr[44]; memset(hdr, 0, 44);
    memcpy(hdr, "RIFF", 4); *(int*)(hdr + 4) = 36 + data;
    memcpy(hdr + 8, "WAVE", 4); memcpy(hdr + 12, "fmt ", 4);
    *(int*)(hdr + 16) = 16; *(short*)(hdr + 20) = 1; *(short*)(hdr + 22) = (short)ch;
    *(int*)(hdr + 24) = sr; *(int*)(hdr + 28) = sr * ch * (bits / 8);
    *(short*)(hdr + 32) = (short)ch; *(short*)(hdr + 34) = (short)bits;
    memcpy(hdr + 36, "data", 4); *(int*)(hdr + 40) = data;
    fwrite(hdr, 1, 44, f);
    for (size_t i = 0; i < n; i++) {
        int32_t vl = (int32_t)(std::max(-1.0f, std::min(1.0f, L[i])) * 8388607.0f);
        int32_t vr = (int32_t)(std::max(-1.0f, std::min(1.0f, R[i])) * 8388607.0f);
        unsigned char b0 = vl & 0xFF, b1 = (vl >> 8) & 0xFF, b2 = (vl >> 16) & 0xFF;
        fwrite(&b0, 1, 1, f); fwrite(&b1, 1, 1, f); fwrite(&b2, 1, 1, f);
        b0 = vr & 0xFF; b1 = (vr >> 8) & 0xFF; b2 = (vr >> 16) & 0xFF;
        fwrite(&b0, 1, 1, f); fwrite(&b1, 1, 1, f); fwrite(&b2, 1, 1, f);
    }
    fclose(f);
    return true;
 }
 static std::vector<float> resample_mono(const std::vector<float>& in, int src_sr, int dst_sr) {
    double frac = (double)dst_sr / src_sr;
    int out_len = (int)(in.size() * frac) + 16;
    std::vector<float> buf(out_len);
    SRC_DATA sd;
    sd.data_in = in.data(); sd.input_frames = (long)in.size();
    sd.data_out = buf.data(); sd.output_frames = out_len;
    sd.src_ratio = frac; sd.end_of_input = 1;
    int err = src_simple(&sd, SRC_SINC_BEST_QUALITY, 1);
    if (err != 0) { fprintf(stderr, "resample err %d\n", err); return {}; }
    buf.resize(sd.output_frames_gen);
    return buf;
 }
 static std::vector<float> resample_stereo(const std::vector<float>& in, int src_sr, int dst_sr) {
    size_t n = in.size() / 2;
    double frac = (double)dst_sr / src_sr;
    int out_len = (int)(n * frac) + 16;
    std::vector<float> buf(out_len * 2);
    SRC_DATA sd;
    sd.data_in = in.data(); sd.input_frames = (long)n;
    sd.data_out = buf.data(); sd.output_frames = out_len;
    sd.src_ratio = frac; sd.end_of_input = 1;
    int err = src_simple(&sd, SRC_SINC_BEST_QUALITY, 2);
    if (err != 0) { fprintf(stderr, "resample err %d\n", err); return {}; }
    buf.resize(sd.output_frames_gen * 2);
    return buf;
 }
 int main(int argc, char** argv) {
    if (argc < 3) {
        fprintf(stderr, "usage: %s in.wav out.wav fc,q,sens[,scale] ...\n", argv[0]);
        return 1;
    }
    std::vector<float> x; int sr, channels;
    if (!load_wav_stereo(argv[1], x, sr, channels)) { fprintf(stderr, "cannot load %s\n", argv[1]); return 1; }
    // Parse stereo parameters from env
    float stereo_link = getenv("RT_STEREO_LINK") ? atof(getenv("RT_STEREO_LINK")) : 1.0f;
    float stereo_balance = getenv("RT_STEREO_BALANCE") ? atof(getenv("RT_STEREO_BALANCE")) : 0.284f;
    float depth = getenv("RT_DEPTH") ? atof(getenv("RT_DEPTH")) : 1.0f;
    float mix = getenv("RT_MIX") ? atof(getenv("RT_MIX")) : 1.0f;
    // Parse bands
    std::vector<DetectorBand> bands;
    for (int i = 3; i < argc; i++) {
        if (!strchr(argv[i], ',')) continue;
        float fc, q, sens, scl = 1.0f;
        if (sscanf(argv[i], "%f,%f,%f,%f", &fc, &q, &sens, &scl) < 3) continue;
        DetectorBand b; b.fc = fc; b.q = q; b.sens = sens * depth; b.level_scale = scl;
        bands.push_back(b);
    }
    if (bands.empty()) bands.push_back({1000.0f, 1.0f, 12.0f * depth});
    auto x48 = resample_stereo(x, sr, 48000);
    if (x48.empty()) return 1;
    size_t n = x48.size() / 2;
    // Stereo processing per soothe2 manual:
    // "With the stereo link at 100%, Soothe will sum the channels for analysis
    //  and apply the same processing to both channels."
    // Use separate processors for L and R to avoid stateful interference.
    SpectralProcessor procL(4096, 1024, 48000.0f);
    SpectralProcessor procR(4096, 1024, 48000.0f);
    procL.setDetectorParams(bands);
    procR.setDetectorParams(bands);
    std::vector<float> L_out(n), R_out(n);
    const size_t BLK = 1 << 16;
    std::vector<float> inb(BLK), outb(BLK);
    // Process L channel
    for (size_t s = 0; s < n; s += BLK) {
        size_t blk = std::min(BLK, n - s);
        memcpy(inb.data(), x48.data() + 2*s, blk * sizeof(float));
        for (size_t i = blk; i < BLK; i++) inb[i] = 0.0f;
        procL.processBlock(inb.data(), outb.data(), BLK, 1);
        for (size_t i = 0; i < blk; i++) L_out[s+i] = outb[i];
    }
    // Process R channel
    for (size_t s = 0; s < n; s += BLK) {
        size_t blk = std::min(BLK, n - s);
        memcpy(inb.data(), x48.data() + 2*s + 1, blk * sizeof(float));
        for (size_t i = blk; i < BLK; i++) inb[i] = 0.0f;
        procR.processBlock(inb.data(), outb.data(), BLK, 1);
        for (size_t i = 0; i < blk; i++) R_out[s+i] = outb[i];
    }
    // Apply balance and mix
    std::vector<float> L_final(n), R_final(n);
    for (size_t i = 0; i < n; i++) {
        float L = x48[2*i];
        float R = x48[2*i+1];
        float L_proc = L_out[i];
        float R_proc = R_out[i];
        float mask_L = (std::abs(L) > 1e-12f) ? L_proc / L : 1.0f;
        float mask_R = (std::abs(R) > 1e-12f) ? R_proc / R : 1.0f;
        // Apply balance: scale reduction for R channel
        float mask_R_bal = 1.0f - (1.0f - mask_R) * stereo_balance;
        // Apply mix
        L_final[i] = L * (1.0f - mix) + L * mask_L * mix;
        R_final[i] = R * (1.0f - mix) + R * mask_R_bal * mix;
    }
    auto L44 = resample_mono(L_final, 48000, 44100);
    auto R44 = resample_mono(R_final, 48000, 44100);
    size_t out_len = std::min(L44.size(), R44.size());
    out_len = std::min(out_len, x.size() / std::max(channels, 1));
    L44.resize(out_len);
    R44.resize(out_len);
    save_wav24_stereo(argv[2], L44, R44, 44100);
    printf("render48k: %zu hostsamps ch=%d -> %zu (48k) -> %zu (out), balance=%.3f depth=%.3f\n",
           x.size(), channels, x48.size(), out_len, stereo_balance, depth);
    return 0;
 }
@@ -0,0 +1,134 @@
 #pragma once
 // Bit-exact transcription of the sincospi "rotor" kernel 0x180184900 (fast path),
 // per-element scalar form, mirroring the exact VEX/FMA op order of big_184900.dis
 // (main loop 0x180184a80-0x1801850d4 / masked tail 0x180185160-0x18018530c).
 //
 // Per input float x the kernel emits two floats:
 //     bufA[i] = sin(x)   (buf1=rdx/arg at kernel entry)
 //     bufB[i] = cos(x)   (buf2=r8 /arg)
 // (verified: tail loop stores sin->[r11+rbx]=bufA via ymm5, cos->[r11+r12]=bufB via ymm2;
 //  main loop stores the 4 lane-group results ymm{7,6,4,1}->bufA, ymm{8,5,3,2}->bufB).
 //
 // The 8-lane AVX body interleaves four independent groups; all ops are per-lane
 // (no cross-lane shuffle), so the scalar loop below reproduces each element exactly.
 //
 // The slow path (0x180186086, entered when any lane |x| >= 10000 or non-finite) and the
 // inf/nan guard helper 0x180189d60 are NOT reproduced: for audio |x| <= 1 they never run.
 // Contract: inputs must satisfy |x| < 10000 (checked in rotor_sincos via nan guard).
 #include <cstdint>
 #include <cstring>
 #include <cmath>
 namespace detkernel {
 // exact fused multiply-add with a single rounding (no re-contraction by the compiler)
 static inline float fma_f(float a, float b, float c) {
    return __builtin_fmaf(a, b, c);
 }
 static inline uint32_t bit_cast_u32(float f) {
    uint32_t u;
    std::memcpy(&u, &f, sizeof(u));
    return u;
 }
 static inline float bit_cast_f(uint32_t u) {
    float f;
    std::memcpy(&f, &u, sizeof(f));
    return f;
 }
 // magic-rounding trick constants (from tables 0x181d1b3x0 / 0x181d1b0c0-0x181d1b300)
 static constexpr uint32_t C_ABS_MASK   = 0x7fffffffu;   // 0x181d1ad40
 static constexpr float    C_INV_PI     = 0.318309873f;  // 0x181d1b340  4b400000? no: 3ea2f983
 static constexpr float    C_MAGIC      = 12582912.0f;   // 0x181d1b380  0x4b400000
 static constexpr float    C_PI_HI      = 3.141592741f;  // 0x181d1b0c0  0x40490fdb
 static constexpr float    C_PI_LO      = -8.742277658e-08f; // 0x181d1b100 0xb3bbbd2e
 static constexpr float    C_RED_CORR   = -3.430249024e-15f; // 0x181d1b140 0xa7772ced (C0)
 static constexpr float    C_SIN_1      = 0.00833306462f;    // 0x181d1b280 0x3c088768 (C1)
 static constexpr float    C_SIN_2      = -1.980916067e-04f; // 0x181d1b2c0 0xb94fb6cf (C2)
 static constexpr float    C_SIN_3      = 2.604164590e-06f;  // 0x181d1b300 0x362ec335 (C3)
 static constexpr float    C_SIN_N6     = -0.166666612f;     // 0x181d1b180 0xbe2aaaa7
 static constexpr uint32_t C_HALF_BITS  = 0x3f000000u;       // 0.5f  0x181d1b400
 static constexpr float    C_ONE        = 1.0f;              // 0x181d1b440
 static constexpr float    C_NOISE      = 10000.0f;          // 0x181d1ad80 (slow-path threshold)
 static constexpr float    C_MASK_DEF   = 0.75f;             // 0x181d1c280 (filled default lanes)
 // rotor_sincos(x, &sin, &cos) per 0x180184900 (elementwise, exact op order)
 static inline void rotor_sincos(float x, float& s, float& c) {
    uint32_t ux;
    std::memcpy(&ux, &x, sizeof(ux));
    // vmovups/ymm load + vandps abs mask + vfmadd231ps
    uint32_t ua = ux & C_ABS_MASK;
    float a;
    std::memcpy(&a, &ua, sizeof(a));
    float t = fma_f(a, C_INV_PI, C_MAGIC);   // t = magic + a*(1/pi)
    float k = t - C_MAGIC;                   // vsubps k = t - magic
    uint32_t tk;
    std::memcpy(&tk, &t, sizeof(tk));
    uint32_t sbt = tk & 1u;                  // vpslld t,31 -> bit0 (parity/round)
    float p = fma_f(-C_PI_HI, k, a);         // vfnmadd231 p = a - pi_hi*k
    p       = fma_f(-C_PI_LO, k, p);         // p -= pi_lo*k
    uint32_t up;
    std::memcpy(&up, &p, sizeof(up));
    uint32_t sbp = up & 0x80000000u;         // vandps -0.0 -> sign bit of p
    uint32_t halfbits = sbp ^ C_HALF_BITS;   // vxorps (-0.0&p) ^ 0.5  -> +-0.5
    float h;
    std::memcpy(&h, &halfbits, sizeof(h));
    float q = k + h;                         // vaddps q = k +- 0.5
    // vfnmadd213ps dest,src1,src2 = -(dest*src1) + src2
    float cc = fma_f(C_RED_CORR, -k, p);     // cc = p - C0*k          (sin arg core)
    float rq = fma_f(-C_PI_HI, q, a);        // rq = a - pi_hi*q
    rq        = fma_f(-C_PI_LO, q, rq);
    float cq = fma_f(C_RED_CORR, -q, rq);    // cq = rq - C0*q          (cos arg core)
    uint32_t sbth = sbt ? 0x80000000u : 0u;  // vpslld fully left -> 0x80000000/0
    uint32_t us1 = bit_cast_u32(cc) ^ sbth;  // arg_sin = cc ^ sbt
    // vxorps(-0.0,sbp)=0x80000000^sbp, then ^sbt -> arg_cos = cq ^ sbt ^ sbp ^ 0x80000000
    uint32_t us2 = bit_cast_u32(cq) ^ sbth ^ sbp ^ 0x80000000u;
    float sarg = bit_cast_f(us1);
    float qarg = bit_cast_f(us2);
    float ss = sarg * sarg;                  // vmulps
    float qs = qarg * qarg;
    float as = fma_f(C_SIN_3, ss, C_SIN_2);  // Horner (vfmadd231 then vfmadd213 chain)
    float aq = fma_f(C_SIN_3, qs, C_SIN_2);
    as = fma_f(as, ss, C_SIN_1);
    aq = fma_f(aq, qs, C_SIN_1);
    as = fma_f(as, ss, C_SIN_N6);
    aq = fma_f(aq, qs, C_SIN_N6);
    float os = ss * as;                      // vmulps
    float oq = qs * aq;
    float sins = fma_f(os, sarg, sarg);      // vfmadd213 sin = os*sarg + sarg
    float sinq = fma_f(oq, qarg, qarg);
    uint32_t usrc = ux & 0x80000000u;        // vandnps ~abs & src -> sign bit of src
    s = bit_cast_f(bit_cast_u32(sins) ^ usrc); // vxorps sin ^ sign(src)
    bool zero = (ux == 0u) || (ux == 0x80000000u); // vcmpeqps src == signbit(src)
    c = zero ? C_ONE : sinq;                  // vblendvps -> 1.0 for zero lanes
 }
 // 8-lane rotor; mirrors the AVX main loop over 32-float chunks implicitly (loop of 8)
 // and supports the masked tail (partial) via `count`. When count < pushed, the masked
 // lanes of the group are handled by the caller with bait; here we simply clamp.
 static inline void rotor_batch(const float* src, float* sin_out, float* cos_out, size_t n) {
    for (size_t i = 0; i < n; ++i) {
        float si, co;
        rotor_sincos(src[i], si, co);
        sin_out[i] = si;
        cos_out[i] = co;
    }
 }
 } // namespace detkernel
@@ -0,0 +1,14 @@
 #pragma once
 // DIVIDE 1803a06a0 tables from .rdata RVA 0x21269c0/0x2126a00 poly 0.207
 // BLOCKMAP:580 vpermps+poly 1c dump
 // extracted via PE .rdata .rdata VA 0x1baa000 raw 0x1ba8400
 #include <array>
 namespace rt_div {
 constexpr std::array<float,32> tbl_1269c0 = {
 127.0f,126.830078f,126.67807f,126.540558f,126.415039f,126.299561f,126.192642f,126.093109f,127.0f,126.830078f,126.67807f,126.540558f,126.415039f,126.299561f,126.192642f,126.093109f,0.0f,-3.12644238e-06f,1.8367532e-06f,1.05200343e-05f,-1.56322119e-06f,-2.65016098e-07f,2.86602835e-06f,2.73532123e-07f,0.0f,-3.12644238e-06f,1.8367532e-06f,1.05200343e-05f,-1.56322119e-06f,-2.65016098e-07f,2.86602835e-06f,2.73532123e-07f};
 constexpr std::array<float,32> tbl_126a00 = {
 0.0f,-3.12644238e-06f,1.8367532e-06f,1.05200343e-05f,-1.56322119e-06f,-2.65016098e-07f,2.86602835e-06f,2.73532123e-07f,0.0f,-3.12644238e-06f,1.8367532e-06f,1.05200343e-05f,-1.56322119e-06f,-2.65016098e-07f,2.86602835e-06f,2.73532123e-07f,0.20751521f,0.20751521f,0.20751521f,0.20751521f,0.20751521f,0.20751521f,0.20751521f,0.20751521f,0.0f,0.0f,0.0f,0.0f,0.0f,0.0f,0.0f,0.0f};
 constexpr std::array<float,9> poly_0 = {
 0.20751521f,-0.24168693f,0.28853500f,-0.36067134f,0.48089835f,0.24026407f,0.05551192f,0.69314718f,1.92590424e-08f};
 }
@@ -0,0 +1,14 @@
 // rt_mask_tables.hpp — runtime mask-apply tables captured from live snapshot
 // (snap_rt.bin, ctx 0x2370040, SR=48000/N=4096). Bit-exact mask chain (FUN_180529fe0).
 // IIR stage: y = A[i]*acc + B[i]*x  (first-order leaky, B = 1 - A).
 #pragma once
 extern const double kIIR_A1[];
 extern const double kIIR_B1[];
 extern const double kIIR_A2[];
 extern const double kIIR_B2[];
 extern const double kIIR_A3[];
 extern const double kIIR_B3[];
 extern const float kWarp[];
 extern const float kBand768[];
 extern const float kPRNGLut[];
@@ -0,0 +1,5 @@
 // Auto-generated from live RT capture /tmp/snap_rt.bin (0x5406c8/0x5406e8).
 // per-bin attack (att) and release (rel) coefficients for the level tracker.
 #pragma once
 extern const float kRTAtt[2049];
 extern const float kRTRel[2049];
@@ -1,33 +1,51 @@
 #include "spectral.hpp"
 #include "fftconv.hpp"
 #include "log2_ln.hpp"
 #include "exp2_tables.hpp"
 #include "exp2.hpp"
 #include <cmath>
 #include <cstring>
 #include <vector>
 #include <cstdlib>
 #include <cstdio>
-SpectralProcessor::SpectralProcessor(size_t nfft, size_t hop)
+SpectralProcessor::SpectralProcessor(size_t nfft, size_t hop, float sample_rate)
    : nfft_(nfft), hop_(hop), frame_count_(0), output_pos_(0),
-      detector_(nfft, 44100.0f) {
+      detector_(nfft, sample_rate) {
-    window_ = new double[nfft_];
+    window_.resize(nfft_);
    computeWindow();
    fft::init_plan(&plan_, static_cast<uint32_t>(std::log2(nfft_)));
-    buf_ = new std::complex<double>[nfft_];
+    buf_.resize(nfft_);
-    tmp_buf_ = new std::complex<double>[nfft_];
+    tmp_buf_.resize(nfft_);
    fir_buf_.resize(nfft_);
    fir_freq_.resize(nfft_);
    overlap_.resize(nfft_, 0.0f);
    mask_.resize(nfft_, 1.0f);
    // Build FIR window: falling half of periodic Hann(4096).
    // Plugin reads window[N/2..N-1] of periodic Hann (rising 0→1).
    fir_window_.resize(nfft_);
    for (size_t i = 0; i < nfft_; i++) {
        fir_window_[i] = 0.5 * (1.0 - std::cos(2.0 * M_PI * i / nfft_));
    }
 }
-SpectralProcessor::~SpectralProcessor() {
+SpectralProcessor::~SpectralProcessor() = default;
    delete[] window_;
    delete[] buf_;
    delete[] tmp_buf_;
 }
-void SpectralProcessor::setDetectorParams(float sharpness, float selectivity, float depth) {
+void SpectralProcessor::setDetectorParams(const std::vector<DetectorBand>& bands) {
-    detector_.setParams(sharpness, selectivity, depth);
+    detector_.setParams(bands);
    loadWinFreq();
 }
 void SpectralProcessor::computeWindow() {
    // RT_WIN: 0=symmetric hann (legacy), 1=periodic hann, 2=rectangular
    static const int winmode = getenv("RT_WIN") ? atoi(getenv("RT_WIN")) : 0;
    for (size_t i = 0; i < nfft_; i++) {
-        window_[i] = 0.5 * (1.0 - std::cos(2.0 * M_PI * i / (nfft_ - 1)));
+        double v;
        if (winmode == 1)      v = 0.5 * (1.0 - std::cos(2.0 * M_PI * i / nfft_));
        else if (winmode == 2) v = 1.0;
        else                   v = 0.5 * (1.0 - std::cos(2.0 * M_PI * i / (nfft_ - 1)));
        window_[i] = v;
    }
 }
@@ -39,20 +57,24 @@ void SpectralProcessor::stftFrame(const float* in, std::complex<double>* out) {
 }
 void SpectralProcessor::istftFrame(std::complex<double>* in, float* out, float* overlap) {
-    memcpy(tmp_buf_, in, nfft_ * sizeof(std::complex<double>));
+    memcpy(tmp_buf_.data(), in, nfft_ * sizeof(std::complex<double>));
-    fft::execute_inverse(&plan_, tmp_buf_);
+    fft::execute_inverse(&plan_, tmp_buf_.data());
    static bool wola_computed = false;
    static float wola_norm = 1.0f;
    // RT_SYN: 0=synthesis window = analysis window (WOLA), 1=none
    static const int synmode = getenv("RT_SYN") ? atoi(getenv("RT_SYN")) : 0;
    if (!wola_computed) {
        double wola_sum = 0.0;
        for (size_t i = 0; i < nfft_; i++) {
-            wola_sum += window_[i] * window_[i];
+            double w = (synmode == 1) ? 1.0 : window_[i];
            wola_sum += window_[i] * w;
        }
        wola_norm = static_cast<float>(wola_sum / hop_);
        wola_computed = true;
    }
    for (size_t i = 0; i < nfft_; i++) {
-        overlap[i] += static_cast<float>(tmp_buf_[i].real() * window_[i]);
+        double w = (synmode == 1) ? 1.0f : window_[i];
        overlap[i] += static_cast<float>(tmp_buf_[i].real() * w);
    }
    for (size_t i = 0; i < hop_; i++) {
        out[i] = overlap[i] / wola_norm;
@@ -65,25 +87,230 @@ void SpectralProcessor::istftFrame(std::complex<double>* in, float* out, float*
    }
 }
 void SpectralProcessor::loadWinFreq() {
    if (win_freq_loaded_) return;
    win_freq_loaded_ = true;
    // Try to load WIN_freq from live capture (handoff/rtwin_freq_44100.npy)
    FILE* f = fopen("handoff/rtwin_freq_44100.npy", "rb");
    if (!f) {
        // Fallback: compute periodic Hann, second half (0.5→1.0 rising)
        win_freq_.resize(nfft_ / 2 + 1);
        for (size_t i = 0; i <= nfft_ / 2; i++) {
            win_freq_[i] = static_cast<float>(0.5 * (1.0 - std::cos(2.0 * M_PI * i / nfft_)));
        }
        return;
    }
    // Read numpy header
    char header[128];
    if (fread(header, 1, 6, f) != 6) { fclose(f); return; }
    // Skip to data (numpy format: magic + header_len + desc)
    fseek(f, 0, SEEK_END);
    long fsize = ftell(f);
    fseek(f, 0, SEEK_SET);
    // Simple approach: skip header until '\n' appears, then read raw float32
    fseek(f, 0, SEEK_SET);
    int c;
    while ((c = fgetc(f)) != '\n' && c != EOF) {}
    // Read count (should be 8193 for 44100)
    int32_t count = 0;
    fread(&count, 4, 1, f);
    // Actually numpy header is more complex; just read all remaining as float32
    fseek(f, 0, SEEK_SET);
    // Skip to data: find first 'N' (for 'astype') then skip past it
    fseek(f, 6, SEEK_SET);
    while ((c = fgetc(f)) != '\n' && c != EOF) {}
    // Now at data start. Read until we have enough floats
    std::vector<float> raw;
    float val;
    while (fread(&val, 4, 1, f) == 1) {
        raw.push_back(val);
    }
    fclose(f);
    if (raw.size() > 0) {
        win_freq_ = raw;
    } else {
        // Fallback
        win_freq_.resize(nfft_ / 2 + 1);
        for (size_t i = 0; i <= nfft_ / 2; i++) {
            win_freq_[i] = static_cast<float>(0.5 * (1.0 - std::cos(2.0 * M_PI * i / nfft_)));
        }
    }
 }
 void SpectralProcessor::buildFirFromMask(const float* mask, std::complex<double>* fir, size_t nbin) {
    // Bit-exact FIR construction pipeline from decompilation (BLOCKMAP 24mm9):
    // 1. design = ln(mask) → negate
    // 2. opA = inv-RFFT (th2180) with buf548
    // 3. fold: bins 1..2047 *= 2.0, bins 2049..4095 = 0
    // 4. opB = fwd-RFFT (th1a90) with buf548
    // 5. EXP: complex polynomial exp with q≈0.80 scaling
    // 6. opC = inv-RFFT (th2180) with buf548
    // 7. zero Nyquist
    // 8. window: falling Hann WIN_freq[2048..4095] (w[1024]=0.5, w[2048]=1.0)
    // 9. opD = fwd-RFFT (th1a90) with buf548
    // 10. normalize: FIR[0]=1.0, FIR[1]=0.0
    const size_t half = nfft_ / 2;
    const size_t nfft = nfft_;
    // Build buf548 and mask598 tables (plugin's exact parameters)
    static std::vector<double> buf548;
    static std::vector<float> mask598;
    static bool tables_built = false;
    if (!tables_built) {
        buf548.resize(nfft);  // N doubles = 2 * N/2 entries
        mask598.resize(nfft / 4);  // N/4 floats
        fft::build_buf548(buf548.data(), nfft);
        fft::build_mask598(mask598.data(), nfft);
        tables_built = true;
    }
    // Step 1: design = ln(mask) and negate (already in real domain)
    // Input is real mask [nbin], convert to real array for RFFT
    std::vector<double> design(nfft, 0.0);
    for (size_t i = 0; i <= half; i++) {
        float m = mask[i];
        if (m > 1e-12f) {
            float ln_m = soothe2::ln_plugin_f32(m);
            design[i] = -static_cast<double>(ln_m);
        } else {
            design[i] = 0.0;
        }
    }
    // Step 2: opA = inv-RFFT (th2180): design (real) → time domain
    // But wait: inv-RFFT takes N/2+1 complex → N real
    // We need to pack design as complex first (im=0)
    std::vector<std::complex<double>> H(half + 1);
    for (size_t i = 0; i <= half; i++) {
        H[i] = std::complex<double>(design[i], 0.0);
    }
    std::vector<double> time_domain(nfft);
    fft::execute_real_inverse_exact(&plan_, H.data(), time_domain.data(), buf548.data(), mask598.data());
 // Step 3: fold - from BLOCKMAP: "FIR[n]=0 (n=0x540534=4096!)"
    // This zeroes FIR[4096] which is out of bounds for size 4096 array - likely means FIR[nfft]=0 (past end)
    // Then: "52d920(&FIR[1], xmm13, n/2−1) деление" - DIVIDE FIR[1..2047]
    // "52db50(&FIR[2049], xmm9, n/2−1)" - multiply/zero FIR[2049..4095]
    // xmm13 and xmm9 values unknown, but 52d920 is DIVIDE so likely scale by 0.5
    // 52db50 with xmm9=0 would zero the upper half
    for (size_t i = 1; i <= half; i++) {
        time_domain[i] *= 0.5;  // DIVIDE by 2 (xmm13 = 0.5?)
    }
    for (size_t i = half + 1; i < nfft; i++) {
        time_domain[i] = 0.0;  // xmm9 = 0 zeros upper half
    }
    // Step 4: opB = fwd-RFFT (th1a90): time_domain (real) → complex
    std::vector<std::complex<double>> freq_domain(half + 1);
    fft::execute_real_forward_exact(&plan_, time_domain.data(), freq_domain.data(), buf548.data(), mask598.data());
    // Step 5: EXP: complex polynomial exp with q≈0.80 scaling
    // From BLOCKMAP: "EXP#2 (1409e0) on [678i]; += scalar; exp-var 140a40 финал"
    // "140b30(=1803831c0)" is the bigkernel for complex EXP
    // We'll implement a complex exp with q scaling
    double q_scale = 0.80;
    for (size_t i = 0; i <= half; i++) {
        double re = freq_domain[i].real();
        double im = freq_domain[i].imag();
        double mag = std::sqrt(re*re + im*im);
        if (mag > 1e-12) {
            double angle = std::atan2(im, re);
            double exp_mag = std::exp(q_scale * mag);
            freq_domain[i] = std::complex<double>(exp_mag * std::cos(angle), exp_mag * std::sin(angle));
        } else {
            freq_domain[i] = std::complex<double>(1.0, 0.0);
        }
    }
    // Step 6: opC = inv-RFFT (th2180): freq_domain → time domain
    std::vector<double> time_domain2(nfft);
    fft::execute_real_inverse_exact(&plan_, freq_domain.data(), time_domain2.data(), buf548.data(), mask598.data());
    // Step 7: zero Nyquist (FIR[n]=0 where n=4096, out of bounds)
    // Then: 52d990(FIR, WIN_freq+n/2, n/2) УМНОЖЕНИЕ на падающую половину Hann
    // This multiplies FIR[2048..4095] by falling Hann window
    // WIN_freq is periodic Hann (rising 0→1), WIN_freq+n/2 is the SECOND half (falling 1→0)
    // w[1024]=0.5, w[2048]=1.0 means:
    // - For i=2048 (offset 0): window = WIN_freq[2048+0] = WIN_freq[2048] = 1.0
    // - For i=3072 (offset 1024): window = WIN_freq[2048+1024] = WIN_freq[3072] = 0.5
    // - For i=4095 (offset 2047): window = WIN_freq[2048+2047] = WIN_freq[4095] = 0.0
    for (size_t i = half; i < nfft; i++) {
        size_t win_idx = half + (i - half);
        if (win_idx < win_freq_.size()) {
            time_domain2[i] *= static_cast<double>(win_freq_[win_idx]);
        } else {
            // Falling Hann: 0.5 * (1.0 + cos(2*pi*i/N))
            double win = 0.5 * (1.0 + std::cos(2.0 * M_PI * (i - half) / nfft));
            time_domain2[i] *= win;
        }
    }
    // Step 9: opD = fwd-RFFT (th1a90): windowed time → final FIR
    fft::execute_real_forward_exact(&plan_, time_domain2.data(), fir, buf548.data(), mask598.data());
    // Step 10: normalize: FIR[0]=1.0, FIR[1]=0.0
    fir[0] = std::complex<double>(1.0, 0.0);
    if (half > 1) {
        fir[1] = std::complex<double>(0.0, 0.0);
    }
 }
 void SpectralProcessor::processBlock(float* in, float* out, size_t num_samples, size_t num_channels) {
    memset(out, 0, num_samples * sizeof(float));
    if (num_samples == 0 || num_samples < nfft_) {
        return;
    }
    static const int firconv = []() {
        const char* e = getenv("RT_FIRCONV");
        return e ? atoi(e) : 0;
    }();
    size_t nframes = (num_samples - nfft_) / hop_ + 1;
    for (size_t f = 0; f < nframes; f++) {
        size_t offset = f * hop_;
        if (offset + nfft_ > num_samples) break;
-        stftFrame(in + offset, buf_);
+        stftFrame(in + offset, buf_.data());
-        detector_.processFrame(buf_, mask_.data());
+        detector_.processFrame(buf_.data(), mask_.data());
-        for (size_t i = 0; i < nfft_; i++) {
+        if (firconv == 3) {
-            buf_[i] *= mask_[i];
+            // RT_FIRCONV=3 (NOTES 24k): plugin application law decoded live:
            //   applied_gain = 1.019 * V^1.8345 per bin (rms 0.0025 dB over
            //   8 drive levels). V = band curve (detector output); here M.
            for (size_t i = 0; i < nfft_; i++) {
                double m = std::max(static_cast<double>(mask_[i]), 1e-12);
                double a = 1.019 * std::pow(m, 1.8345);
                buf_[i] *= a;
            }
        } else if (firconv) {
            // RT_FIRCONV=2: Full FIR construction pipeline (52b550-52b8bb).
            // mask → reciprocal (1/mask) → window → normalize → complex multiply.
            // This replicates the plugin's FFT-conv FIR design path.
            buildFirFromMask(mask_.data(), fir_freq_.data(), nfft_);
            // Complex multiply FIR × audio spectrum
            for (size_t i = 0; i < nfft_; i++) {
                buf_[i] *= fir_freq_[i];
            }
        } else if (firconv == 1) {
            // RT_FIRCONV=1: Simple frequency-domain mask multiply (legacy).
            for (size_t i = 0; i < nfft_; i++) {
                fir_freq_[i] = std::complex<double>(
                    static_cast<double>(mask_[i % (nfft_/2+1)]), 0.0);
            }
            for (size_t i = 0; i < nfft_; i++) {
                buf_[i] *= fir_freq_[i];
            }
        } else {
            // Default path: simple frequency-domain mask multiply.
            for (size_t i = 0; i < nfft_; i++) {
                buf_[i] *= mask_[i];
            }
        }
-        istftFrame(buf_, out + offset, overlap_.data());
+        istftFrame(buf_.data(), out + offset, overlap_.data());
    }
 }
@@ -4,33 +4,47 @@
 #include <complex>
 #include <vector>
 #include "fft.hpp"
-#include "detect.hpp"
+#include "framed_model.hpp"
 constexpr size_t DEFAULT_NFFT = 2048;
 constexpr size_t DEFAULT_HOP = 512;
 class SpectralProcessor {
 public:
-    SpectralProcessor(size_t nfft = DEFAULT_NFFT, size_t hop = DEFAULT_HOP);
+    SpectralProcessor(size_t nfft = DEFAULT_NFFT, size_t hop = DEFAULT_HOP,
                      float sample_rate = 44100.0f);
    ~SpectralProcessor();
-    void setDetectorParams(float sharpness, float selectivity, float depth);
+    void setDetectorParams(const std::vector<DetectorBand>& bands);
    void processBlock(float* in, float* out, size_t num_samples, size_t num_channels = 1);
 private:
    size_t nfft_;
    size_t hop_;
-    double* window_;
+    std::vector<double> window_;
    FFTPlan plan_;
-    std::complex<double>* buf_;
+    std::vector<std::complex<double>> buf_;
-    std::complex<double>* tmp_buf_;
+    std::vector<std::complex<double>> tmp_buf_;
    std::vector<std::complex<double>> fir_buf_;
    std::vector<std::complex<double>> fir_freq_;
    std::vector<double> fir_window_;
    std::vector<float> overlap_;
    std::vector<float> mask_;
-    Detector detector_;
+    FramedDetector detector_;
    size_t frame_count_;
    size_t output_pos_;
    void computeWindow();
    void stftFrame(const float* in, std::complex<double>* out);
    void istftFrame(std::complex<double>* in, float* out, float* overlap);
    // FIR construction from detector mask (52b550-52b8bb pipeline):
    // mask → log → sign-invert → EXP → twiddle ops → window → normalize
    // Produces frequency-domain FIR kernel for complex multiply application.
    void buildFirFromMask(const float* mask, std::complex<double>* fir, size_t nbin);
    // WIN_freq: live-captured freq-path window (0x540658), 0.5→1.0
    std::vector<float> win_freq_;
    bool win_freq_loaded_ = false;
    void loadWinFreq();
 };
@@ -0,0 +1,59 @@
 #include <cstdio>
 #include <cmath>
 #include <cstring>
 #include <vector>
 #include "dsp_ctx.hpp"
 #include "tables_data.hpp"
 int main() {
    const size_t N = 8193;
    bool ok = true;
    auto report = [&](const char* name, double v, double expect, double tol) {
        bool p = std::fabs(v - expect) < tol;
        ok = ok && p;
        std::printf("  %-28s %12.6f  (%s)\n", name, v, p ? "ok" : "MISMATCH");
    };
    std::printf("dsp_ctx tables synthetic check (from runtime capture npy)\n");
    std::printf("window WIN_WINDOW[%zu]\n", WIN_WINDOW_COUNT);
    report("win[0]", WIN_WINDOW[0], 0.5, 1e-6);
    report("win[N-1]", WIN_WINDOW[N - 1], 1.0, 1e-6);
    double mono = 0.0;
    for (size_t i = 0; i < N; i++) mono += WIN_WINDOW[i];
    report("win sum", mono, 0.0, 1e9);
    std::printf("weights: WA[%zu] WB[%zu] WC[%zu] WD[%zu]\n",
        WTA_WEIGHT_COUNT, WTB_WEIGHT_COUNT, WTC_WEIGHT_COUNT, WTD_WEIGHT_COUNT);
    // Active half is bins 0..2048 (4096-point FFT half); mirror half is zeroed.
    const size_t half = 2049;
    report("WA active tail[2048]", WTA_WEIGHT[2048], 0.125666, 1e-4);
    report("WA mirror zero", WTA_WEIGHT[4096], 0.0, 1e-9);
    report("WB mirror zero", WTB_WEIGHT[8192], 0.0, 1e-9);
    double m = 0.0;
    for (size_t i = 0; i < half; i++) {
        double err = std::fabs(WTA_WEIGHT[i] + WTB_WEIGHT[i] - 1.0);
        if (err > m) m = err;
    }
    report("max|WA+WB-1| (0..2048)", m, 0.0, 1e-5);
    m = 0.0;
    for (size_t i = 0; i < half; i++) {
        double err = std::fabs(WTC_WEIGHT[i] + WTD_WEIGHT[i] - 1.0);
        if (err > m) m = err;
    }
    report("max|WC+WD-1| (0..2048)", m, 0.0, 1e-5);
    report("WA[0]", WTA_WEIGHT[0], 0.596076, 1e-4);
    report("WA[1024]", WTA_WEIGHT[1024], 0.168067, 1e-5);
    std::printf("freq axis FREQAXIS[%zu] @48000 internal (step=48000/4098)\n", WIN_FREQAXIS_COUNT);
    report("fa[0]", WIN_FREQAXIS[0], 0.0, 1e-6);
    report("fa[1]-fa[0]", WIN_FREQAXIS[1] - WIN_FREQAXIS[0], 11.713, 0.01);
    report("fa[N-1]", WIN_FREQAXIS[WIN_FREQAXIS_COUNT - 1], 23976.574, 0.1);
    report("freq_at(50)", dsp_ctx::freq_at(50.0f), 50.0f * 11.713f, 1.0);
    std::printf("dsp_ctx::hz_of_bin(1000,4096) = %.2f\n",
        static_cast<double>(dsp_ctx::hz_of_bin(1000, 4096)));
    std::printf(ok ? "ALL OK\n" : "FAILURES\n");
    return ok ? 0 : 1;
 }
@@ -0,0 +1,115 @@
 #include "twin.hpp"
 #include <cmath>
 #include <cstring>
 #include <xmmintrin.h>
 #include "rotor_kernel.hpp"
 namespace detkernel {
 // ---------------------------------------------------------------------------
 // Generator FUN_180533ec0 (double), then cvtpd2ps packing as in the twin prologue.
 // ---------------------------------------------------------------------------
 twin_coeff build_twin_coeff(double fs_total, double freq, double q, float sens_lin) {
    float f1 = std::sqrt(sens_lin);
    if (f1 <= 0.0f) f1 = 0.0f;
    const double f1d = static_cast<double>(f1);
    const double w0 = (freq < 2.0 ? 2.0 : freq) * 6.283185307179586 / fs_total;
    const double s  = std::sin(w0);
    const double c2 = std::cos(w0) * -2.0;
    const double p  = (s * 0.5) / q;
    const double a0 = 1.0 + p * f1d;
    const double a2 = 1.0 - p * f1d;
    const double b0 = 1.0 + p / f1d;
    const double b2 = 1.0 - p / f1d;
    twin_coeff c;
    c.A[0] = static_cast<float>(a0);
    c.A[1] = static_cast<float>(c2);
    c.A[2] = static_cast<float>(a2);
    c.B[0] = static_cast<float>(b0);
    c.B[1] = static_cast<float>(c2);
    c.B[2] = static_cast<float>(b2);
    return c;
 }
 // ---------------------------------------------------------------------------
 // 0x181a77520: complex division  num/den = (num*conj(den)) * ref
 //   per-lane: shufps-0x88/0xdd -> |den|^2; mulps bith; rcpps + 1 Newton step.
 // ---------------------------------------------------------------------------
 void cplx_div_exact(const cplxf& num, const cplxf& den, cplxf& out) {
    const float ar = den.re, ai = den.im;
    const float br = num.re, bi = num.im;
    const float ar2 = ar * ar;
    const float ai2 = ai * ai;
    const float den2 = ar2 + ai2;
    const float re = ar * br + ai * bi;   // Re{num*conj(den)}
    const float im = ar * bi - ai * br;   // Im{num*conj(den)}
    if (den2 == 0.0f) {
        const float qnan = 0.0f / 0.0f;
        out.re = qnan;
        out.im = qnan;
        return;
    }
    const float r0  = _mm_cvtss_f32(_mm_rcp_ss(_mm_set_ss(den2)));
    const float t1  = den2 * r0;
    const float t2  = 2.0f - t1;
    const float ref = r0 * t2;
    out.re = re * ref;
    out.im = im * ref;
 }
 // ---------------------------------------------------------------------------
 // 0x18000ad60 scalar complex multiply (body @0x18000ae00):
 //   t0=bi*ai; t1=bi*ar; re=fma(br,ar,-t0); im=fma(br,ai,+t1)
 // ---------------------------------------------------------------------------
 void cplx_mul_exact(const cplxf& a, const cplxf& b, cplxf& out) {
    const float ar = a.re, ai = a.im;
    const float br = b.re, bi = b.im;
    const float t0 = bi * ai;   // mulps
    const float t1 = bi * ar;   // mulps
    out.re = fma_f(br, ar, -t0);  // vfmaddsub213ps lane0 (subtract)
    out.im = fma_f(br, ai, t1);   // lane1 (add)
 }
 // ---------------------------------------------------------------------------
 // 0x180535880 hot loop: seed A/B into acc, 2 Horner FMA stages, cplx-div, x2.
 // `z` carries exp(+i*theta_k); the twin conjugates before use (0x1800018b0).
 // ---------------------------------------------------------------------------
 void twin_apply(const twin_coeff& c, const cplxf* z, size_t n, cplxf* out) {
    for (size_t i = 0; i < n; ++i) {
        // conj(z1) = conj(exp(+i*theta))  (0x1800018b0 negates imag)
        const cplxf z1 = { z[i].re, -z[i].im };
        cplxf z2;
        cplx_mul_exact(z1, z1, z2);   // conj(z1)^2
        // A accumulator (vfmadd213ss per scalar lane)
        cplxf accA = { c.A[0], 0.0f };
        accA.re = fma_f(c.A[1], z1.re, accA.re);
        accA.im = fma_f(c.A[1], z1.im, accA.im);
        accA.re = fma_f(c.A[2], z2.re, accA.re);
        accA.im = fma_f(c.A[2], z2.im, accA.im);
        // B accumulator
        cplxf accB = { c.B[0], 0.0f };
        accB.re = fma_f(c.B[1], z1.re, accB.re);
        accB.im = fma_f(c.B[1], z1.im, accB.im);
        accB.re = fma_f(c.B[2], z2.re, accB.re);
        accB.im = fma_f(c.B[2], z2.im, accB.im);
        cplx_div_exact(accB, accA, out[i]);   // B/A
        out[i].re *= 2.0f;                    // 2*B/A (caller-side scale, exact)
        out[i].im *= 2.0f;
    }
 }
 }  // namespace detkernel
@@ -0,0 +1,49 @@
 #pragma once
 #include <cstddef>
 #include <cstdint>
 // Bit-exact transcription of the soothe2 "twin" resonance filter 0x180535880
 // (float sibling of the resonator 2nd-order detector core, NLS .sdk plugin).
 //
 // Pipeline per complex bin (all float32, op-order faithful to the disassembly):
 //   z1 = conj(e^{i*theta})                 (rotor -> interleave, then 0x1800018b0 conj)
 //   z2 = cplx_mul(z1, z1)                  (0x18000ad60 scalar body @0x18000ae00)
 //   accA = A0; accA = fma(A1, z1, accA); accA = fma(A2, z2, accA)   (vfmadd213ss)
 //   accB = B0; accB = fma(B1, z1, accB); accB = fma(B2, z2, accB)
 //   out  = 2 * cplx_div(accB, accA)        (0x181a77520: rcpps + 1 Newton step)
 //
 // Coefficients come from FUN_180533ec0 (double pipeline) + cvtpd2ps packing
 // as done in the twin prologue (stack slots +0x28..+0x40).
 //
 // Constants locked in Phase 1:
 //   A0=B0=1+d, A1=B1=-2*cos(w0), A2=1-d, B2=1-d2,
 //   d  = p*sqrtf(param_5), d2 = p/sqrtf(param_5),
 //   p  = sin(w0)*0.5/Q,  w0 = max(freq,2.0)*2*pi/fs_total.
 //   param_5 = 10^(sens_stored/20) with sens_stored~=24.65 dB (host-scaled ~=2*XML 12.0).
 namespace detkernel {
 struct cplxf {
    float re, im;
 };
 struct twin_coeff {
    float A[3];  // A0,A1,A2  (float32 after cvtpd2ps)
    float B[3];  // B0,B1,B2
 };
 // FUN_180533ec0 coefficients, packed to float32 like the twin prologue.
 // sens_lin = param_5 (linear, before sqrtf) e.g. 10^(24.65/20).
 twin_coeff build_twin_coeff(double fs_total, double freq, double q, float sens_lin);
 // 0x181a77520 complex division, scalar form (rcpps + Newton, NaN guard).
 void cplx_div_exact(const cplxf& num, const cplxf& den, cplxf& out);
 // 0x18000ad60 scalar complex multiply (vfmaddsub213ps form).
 void cplx_mul_exact(const cplxf& a, const cplxf& b, cplxf& out);
 // Full twin evaluation for `n` bins. `z` carries unit-magnitude twiddles
 // exp(+i*theta_k) (the function applies the conjugate itself).
 void twin_apply(const twin_coeff& c, const cplxf* z, size_t n, cplxf* out);
 }  // namespace detkernel
@@ -0,0 +1,55 @@
 #include <cstdio>
 #include <cmath>
 #include "framed_model.hpp"
 // Unit check for the VLAW detector law (BLOCKMAP:314 softplus proxy):
 //   cut  = alpha * ln1p(lvl/beta) + c [+ delta]
 //   mask = 10^(-cut/20)
 // Reference values hand-computed from the dual-calibrated constants
 // (alpha=3.2193, beta=0.4927, c=0.5423, delta=6.9177 — README.md:26).
 int main() {
    int fail = 0;
    // --- law monotonicity: higher level -> stronger cut -> smaller mask ---
    double m0 = vlaw_mask(0.01, 3.2193, 0.4927, 0.5423, 0.0);
    double m1 = vlaw_mask(1.0,  3.2193, 0.4927, 0.5423, 0.0);
    double m2 = vlaw_mask(10.0, 3.2193, 0.4927, 0.5423, 0.0);
    bool mono = (m0 > m1) && (m1 > m2);
    std::printf("vlaw monotonic: m(0.01)=%.4f m(1)=%.4f m(10)=%.4f (%s)\n",
                m0, m1, m2, mono ? "OK" : "MISMATCH");
    if (!mono) fail = 1;
    // --- zero level: cut = c => mask = 10^(-c/20) ---
    double mz = vlaw_mask(0.0, 3.2193, 0.4927, 0.5423, 0.0);
    double ez = std::pow(10.0, -0.5423 / 20.0);
    bool zok = std::fabs(mz - ez) < 1e-9;
    std::printf("vlaw zero-level: mask=%.6f expect=%.6f (%s)\n",
                mz, ez, zok ? "OK" : "MISMATCH");
    if (!zok) fail = 1;
    // --- delta branch adds cut -> deeper mask ---
    double md = vlaw_mask(1.0, 3.2193, 0.4927, 0.5423, 6.9177);
    bool dok = md < m1;
    std::printf("vlaw delta: mask+delta=%.4f < %.4f (%s)\n",
                md, m1, dok ? "OK" : "MISMATCH");
    if (!dok) fail = 1;
    // --- numeric reference: lvl=1.0, dual params ---
    // cut = 3.2193 * ln(1 + 1/0.4927) + 0.5423
    double cut_ref = 3.2193 * std::log1p(1.0 / 0.4927) + 0.5423;
    double mref = std::pow(10.0, -cut_ref / 20.0);
    bool rok = std::fabs(m1 - mref) < 1e-9;
    std::printf("vlaw ref: mask=%.6f expect=%.6f cut=%.4f (%s)\n",
                m1, mref, cut_ref, rok ? "OK" : "MISMATCH");
    if (!rok) fail = 1;
    // --- comb neutrality: alpha=0.05 beta=5.0 c=0 -> mask ~ 1 for lvl=0 ---
    double mc = vlaw_mask(0.0, 0.05, 5.0, 0.0, 0.0);
    bool cok = std::fabs(mc - 1.0) < 1e-9;
    std::printf("vlaw comb-neutral: mask(0)=%.6f expect=1.0 (%s)\n",
                mc, cok ? "OK" : "MISMATCH");
    if (!cok) fail = 1;
    std::printf("vlaw_check %s\n", fail ? "FAIL" : "PASS");
    return fail;
 }
@@ -0,0 +1,82 @@
 #include "vlog.hpp"
 #include <cmath>
 #include <cstring>
 namespace vlog {
 namespace {
 // Float bit patterns used by the AVX kernel's integer lane ops.
 constexpr uint32_t C_2_OVER_3_BITS = 0x3f2aaaab; // bits of 2/3f (0.6666666865f)
 constexpr uint32_t C_MANTISSA_MASK = 0x007fffff; // 2^23 - 1 (low 23 mantissa bits)
 // ln(2) as a single float constant (0x3f317218).
 constexpr float C_LN2 = 0.6931471824645996f;
 // Minimax polynomial coefficients for ln(1+x) on the reduced interval
 // x in [-1/3, 1/3), read straight from the binary:
 //   ln(1+x) ~= x + x^2 * (c7 + c6*x + c5*x^2 + c4*x^3 + c3*x^4 + c2*x^5 + c1*x^6)
 constexpr float C_P1 = -0.15177205204963684f; // 0x181f82040
 constexpr float C_P2 =  0.16964881122112274f; // 0x181f82020
 constexpr float C_P3 = -0.16462457180023193f; // 0x181f82000
 constexpr float C_P4 =  0.19822503626346588f; // 0x181f81fe0
 constexpr float C_P5 = -0.25004664063453674f; // 0x181f81fc0
 constexpr float C_P6 =  0.33336564898490906f; // 0x181f81fa0
 constexpr float C_P7 = -0.5f;                 // 0x181f81f80
 inline float fma_f(float a, float b, float c) { return __builtin_fmaf(a, b, c); }
 inline uint32_t bit_u32(float f) { uint32_t u; std::memcpy(&u, &f, sizeof u); return u; }
 inline float    bit_f32(uint32_t u) { float f; std::memcpy(&f, &u, sizeof f); return f; }
 // Fast path: input is a positive normal float -> ln(x).
 //
 // Range reduction via the "2/3" magic (equivalent to the kernel's
 // vpsubd/vpsrad 0x17/vpand/vpaddd): the mantissa is folded into
 // m in [2/3, 4/3) and the integer exponent e is recovered, so that
 //   y = m * 2^e   =>   ln(y) = e*ln2 + ln(m),
 // with ln(m) = ln(1+x), x = m-1 in [-1/3, 1/3), from the minimax polynomial.
 inline float ln_fast(float y) {
    uint32_t b  = bit_u32(y);
    uint32_t t  = b - C_2_OVER_3_BITS;                      // vpsubd (wrapping)
    int32_t  e  = static_cast<int32_t>(t) >> 23;            // vpsrad 0x17 (exponent)
    uint32_t mb = (t & C_MANTISSA_MASK) + C_2_OVER_3_BITS;  // vpand + vpaddd
    float m = bit_f32(mb);
    float x = m - 1.0f;                                     // vsubps  (x in [-1/3, 1/3))
    float e_ = static_cast<float>(e);                       // vcvtdq2ps
    // Horner evaluation (mirrors the vfmadd231ps/vfmadd213ps chain).
    float p = C_P2;
    p = fma_f(x, C_P1, p);
    p = fma_f(p, x, C_P3);
    p = fma_f(p, x, C_P4);
    p = fma_f(p, x, C_P5);
    p = fma_f(p, x, C_P6);
    p = fma_f(p, x, C_P7);
    float q = x * p;          // x*P(x)
    q = fma_f(q, x, x);       // x^2*P(x) + x  ~= ln(1+x) = ln(m)
    return fma_f(e_, C_LN2, q); // e*ln2 + ln(m) == ln(y)
 }
 } // namespace
 void log_f32(const float* src, float* dst, uint32_t n) {
    for (uint32_t i = 0; i < n; ++i) {
        float y = src[i];
        uint32_t b = bit_u32(y);
        // Fast path iff the lane is a positive normal float:
        // (int32)(b + 0x00800000) >= 0x01000000  <=>  b in [0x00800000, 0x7f7fffff].
        if (b >= 0x00800000u && b <= 0x7f7fffffu) {
            dst[i] = ln_fast(y);
        } else {
            // Slow path: zero/denormal/negative/Inf/NaN. The kernel dispatches to
            // a scalar double-precision Cody-Waite ln (0x1802a2fc0); std::log is the
            // structurally equivalent reference for these edge inputs.
            dst[i] = static_cast<float>(std::log(static_cast<double>(y)));
        }
    }
 }
 } // namespace vlog
@@ -0,0 +1,20 @@
 #pragma once
 #include <cstdint>
 namespace vlog {
 // Scalar transcription of the vectorized natural-logarithm kernel at 0x1802a24c0
 // (soothe2 VST3, MSVC x86-64 AVX2, extracted from /tmp/snap_rt.bin).
 //
 // The kernel computes  dst[i] = ln(src[i])  elementwise in single precision,
 // using its own minimax polynomial + ln(2) range reduction driven by float
 // bit-manipulation (the "2/3" magic exponent/mantissa split). A scalar
 // double-precision Cody-Waite ln handles the slow path for special values.
 //
 // NOTE: despite the file name, this kernel is a natural logarithm, NOT an FFT
 // butterfly. It sits in the spectral (log-magnitude) processing path, not the
 // complex FFT transform. The signature below mirrors the actual code: float,
 // out-of-place, real (src != dst is allowed).
 void log_f32(const float* src, float* dst, uint32_t n);
 } // namespace vlog
@@ -0,0 +1,63 @@
 #include "vlog.hpp"
 #include <cmath>
 #include <cstdio>
 #include <cstdint>
 #include <vector>
 // Checks the 0x1802a24c0 transcription against the reference natural logarithm.
 //
 // The kernel is a vectorized single-precision ln(x), so the reference is
 // std::log (double) evaluated on the same float inputs. The reported metric is
 // the max relative error over well-conditioned points (|ln(x)| > 1e-6); for
 // inputs where ln(x) ~ 0 (x ~ 1) an absolute error is reported instead.
 // A float minimax polynomial reaches ~1-2 ulp, so a 1e-6 relative gate is the
 // right tolerance (a double FFT-style 1e-9 gate would be unreachable for float).
 static double rel_err(double a, double b) {
    double denom = std::abs(b) > 1e-6 ? std::abs(b) : 1.0;
    return std::abs(a - b) / denom;
 }
 int main() {
    const uint32_t N = 8192;
    std::vector<float> src(N), dst(N);
    // Sweep a wide dynamic range plus a fine neighbourhood around 1.0.
    uint32_t i = 0;
    for (int k = -30; k <= 30 && i < N; ++k) {
        double v = std::pow(10.0, double(k) * 0.5);
        src[i++] = static_cast<float>(v);
    }
    for (int k = -60; k <= 60 && i < N; ++k) {
        src[i++] = static_cast<float>(1.0 + double(k) * 1e-3);
    }
    for (int k = 0; k < 1024 && i < N; ++k) {
        src[i++] = static_cast<float>(double(k + 1) / 1024.0);
    }
    while (i < N) src[i++] = static_cast<float>(i);
    vlog::log_f32(src.data(), dst.data(), N);
    double max_rel = 0.0, max_abs = 0.0;
    uint32_t rel_i = 0, abs_i = 0;
    for (uint32_t j = 0; j < N; ++j) {
        double ref  = std::log(static_cast<double>(src[j]));
        double mine = static_cast<double>(dst[j]);
        double e    = std::abs(mine - ref);
        if (e > max_abs) { max_abs = e; abs_i = j; }
        double r = rel_err(mine, ref);
        if (r > max_rel) { max_rel = r; rel_i = j; }
    }
    std::printf("n = %u\n", N);
    std::printf("max relative error = %.6e  (at src=%.9g, got %.12g, ref %.12g)\n",
                max_rel, src[rel_i], dst[rel_i], std::log(static_cast<double>(src[rel_i])));
    std::printf("max absolute error = %.6e  (at src=%.9g)\n", max_abs, src[abs_i]);
    if (max_rel < 1e-6) {
        std::printf("ALL OK\n");
        return 0;
    }
    std::printf("FAILED\n");
    return 1;
 }
@@ -0,0 +1,23 @@
 -- dump_params.lua : enumerate soothe2 FX params (name, raw, formatted) to file
 local out = io.open("/tmp/opencode/fxparams.txt", "w")
 local tr = reaper.GetTrack(0, 0)
 if tr == nil then
  out:write("NO TRACK\n"); out:close(); return
 end
 local nfx = reaper.TrackFX_GetCount(tr)
 out:write(string.format("nfx=%d\n", nfx))
 for fxi = 0, nfx - 1 do
  local rv, fxname = reaper.TrackFX_GetFXName(tr, fxi, "")
  out:write(string.format("FX %d: %s\n", fxi, fxname))
  local np = reaper.TrackFX_GetNumParams(tr, fxi)
  for p = 0, np - 1 do
    local _, pname = reaper.TrackFX_GetParamName(tr, fxi, p, "")
    local val, minv, maxv = reaper.TrackFX_GetParam(tr, fxi, p)
    local _, fmt = reaper.TrackFX_GetFormattedParamValue(tr, fxi, p, "")
    out:write(string.format("%d\t%s\traw=%.6f\t[%.3f..%.3f]\tfmt=%s\n", p, pname, val, minv, maxv, fmt))
  end
 end
 out:close()
 local t0 = reaper.time_precise()
 while reaper.time_precise() - t0 < 2 do reaper.defer(function() end) end
 reaper.Main_OnCommand(40004, 0) -- File: Quit REAPER
@@ -0,0 +1,33 @@
 import struct
 data=open('/tmp/snap_all.bin','rb').read()
 i=0; regs=[]
 while i+16<=len(data):
    lo,sz,=struct.unpack_from('<QQ',data,i)
    regs.append((lo,sz,data[i+16:i+16+sz]))
    i+=16+sz
 def readabs(addr,n):
    for lo,sz,body in regs:
        if lo<=addr<lo+sz and addr-lo+n<=sz:
            return body[addr-lo:addr-lo+n]
    return None
 known={0x2a72600,0x2111140,0x2cd0fc0,0x29f2280,0x29fa300,0x2a02340,0x2a0a3c0,0x2a12400}
 cands=[]
 t0=len(regs)
 for ri,(lo,sz,body) in enumerate(regs):
    if sz<0x541000: continue
    # check every 8-aligned offset for the +0x540658 ptr direct
    for off in range(0, sz-0x540660, 8):
        b=body[off+0x540658:off+0x540660]
        if len(b)<8: break
        p=struct.unpack_from('<Q',b)[0]
        if p in known:
            base=lo+off
            u24=readabs(base+0x24,4); u28=readabs(base+0x28,4)
            f40=readabs(base+0x40,0x20); f58=readabs(base+0x58,0x20)
            print('cand base=0x%x reg%d p=0x%x u24=%s u28=%s 40=%s'%(
                base,ri,p,
                struct.unpack('<f',u24)[0] if u24 else None,
                struct.unpack('<f',u28)[0] if u28 else None,
                [round(x,4) for x in struct.unpack('<4d',f40[:32])] if f40 else None))
            cands.append(base)
 print('total cand',len(cands))
@@ -0,0 +1,792 @@
 # BLOCK MAP: FUN_180529fe0 (полная разметка по raw asm, 22v)
 Источник: `handoff/nls_dasm/f529fe0_full.dis` (soothe_mem.bin, база 0x180000000,
 диапазон 529c60–52ba00; старый f529fe0.dis был ОБРЕЗАН на 52a813).
 Регистры: r13 = индекс полосы, r12 = state-ptr ctx+0x440518, esi/r8d = nbin,
 rbx = nbands ([rsp+0x30]), rdi = ctx. Флаг 0x5408b8 выбирает float/double
 вариант thunk-операций (семантика пар идентична).
 ## Исправление адресации (главное)
 «Скалярные» буферы NOTES 21b — МАССИВЫ пер-полосных векторов (шаг 16 байт):
 - bands[i] = [ctx+0x540678+i·16] — кривые полос
 - acc[i]   = [ctx+0x5407c8+i·16] — ПЕР-ПОЛОСНЫЙ аккумулятор combine
 - track[i] = [ctx+0x540768+i·16] — трек основного цикла
 ## Thunk-таблица (ILT-стаб 180001xxx -> jmp [table + idx*8], idx @1826159a0)
 Резолв при live idx=4; семантика из тел + NOTES 21b:
 | стаб float | стаб double | impl | op |
 |---|---|---|---|
 | 0x180001f10 | 0x180001c70 | dc40 / **8d60** | dst = B − A (sub) |
 | 0x180001fa0 | 0x180001940 | ee20→487a0 / **3c40** | fma att/rel половин |
 | 0x180001d60 | —(встречен в 1b80-паре?) | — / **5a20** | dst += B (add) |
 | 0x180001970 | — | **3f40** | sub (float) |
 | 0x1800019a0 | — | **4200** | mul scalar? (делегирует 181a63fe0) |
 | 0x180001a00 | — | **4720** | mul |
 | 0x180001a60 | — | **5160** | sub DOUBLE |
 | 0x180001850 | — | **25e0** | add DOUBLE |
 | 0x180002000 | 0x180001c40 | ? | axpy-класс (band ⊕ track) |
 | 0x180002060 | 0x180002120 | 10860/11940 | transform со скаляром xmm3 |
 | 0x180002270 | 0x1800022a0 | 14c40/15060 | transform со скаляром xmm0/xmm1 |
 | 0x180002030 | 0x180001d30 | ? | transform float/double |
 | in-place кернелы | | 140950/140980/1409b0/1409e0/140a40/140ad0/140b00/140b60 | bigkernel-семейство (exp2/mask) |
 ## Карта блоков
 ### Пролог
 - `52a03a–52a396`: PRNG-пролог. LCG state @ctx+0x2404e0, шаги +0x3cdca,
  +0x140236, +0x10d56, +0xdf6b6, +0xa8c5e, +0x72916; LUT ptr @ctx+0x5408b0;
  константы 262b704/262b5c8(int)/262b700(int)/24c3c58. Результат: индексы
  iVar7/iVar8 ([rsp+0x130]/счётчики) и стартовая позиция цикла.
  Live-поведение залочено раньше (fVar30=1).
 - `52a397`: r14 = &bands[0].
 ### Pre-combine #1 (52a397–52a421)
 - `52dbc0(f6f8_data, bands[0], nbin)` — copy bands[0] → f6f8.
 - цикл i=1..nbands−1 по массиву @0x540688 (= &bands[1]!): transform(f6f8, bands[i])
  через 20f0(float)/1850(double=25e0 ADD). ⇒ **f6f8 = Σ_{i≥1} bands[i]** (или min/max
  — точный op 20f0 не залочен, кандидат ADD по double-паре!).
 ### Нормировка + mix-вес (52a421–52a458)
 - `52d920(f6f8, xmm11/[rsp+0x148](int), nbin)` — f6f8 /= K.
 - `xmm7 = powf([ctx+0x2c], xmm13)` — mix^p.
 ### Pre-combine #2 (52a45e–52a4f6) по всем полосам
 - band[i] *= (xmm11 − xmm7)  [thunk 2030/1d30]
 - f6f8 ⊕= ... с весом xmm7   [thunk 1fd0/2150]
 ⇒ взвешенное смешение кривых полос ДО основного цикла.
 ### Основной цикл по полосам (52a4fb..52b3c7, тело с 52a580)
 На каждую полосу i (r13):
 1. `scale` (52a583–52a5c4): s = xmm11·[0x540870]; если флаг 0x5408b8:
   s ← exp(PRNG)-ветка; затем ·[0x54088c]. Читается track_i (@0x540768[i]) в [rsp+0x40].
 2. `transform(bands[i], s)` (52a5cc–52a607): bands[i] *= s  [2030/1d30].
 3. флаг-оп (52a608–52a645): transform(bands[i], bands[i]) через 140980/1409b0
   (bigkernel, вероятно exp2-кернел) — только при флаге.
 4. `52d650(state@0x440518, f6f8, bands[i], nbin)` (52a646–52a658) — БИДИР-IIR #1:
   band → f6f8 (double, коэф down@0x440528/up@0x4c0528, acc@0x540528, reset внутри).
 5. Инлайн БИДИР-IIR #2 (52a65d–52a85a, только при флаге; иначе прыжок на 6):
   вход f6f8 → выход bands[i] (тот же state 0x440518, reset отдельно).
 6. `transform(f6f8 ⊕ bands[i])` (52a85b–52a896) [1970/1a60 = 3f40 SUB float /
   5160 SUB double]: f6f8 −= bands[i]? (аргументы rcx=f6f8, rdx=band).
 7. **БИДИР-IIR #3** (52a897–52aa6e, инлайн): state base 0x3404f8
   (down@0x2404f8, up@0x2c04f8, acc@0x3404f8), длина из поля ctx+0x2404e8,
   IN-PLACE по bands[i]. Reset каждый вызов.
 8. `op(bands[i], скаляр xmm9=0?)` (52aa6e–52aaaf) [2060/2120].
 9. Ветка флага==0 (52aabc–52ab90):
   - op(vec@0x540698, f6f8, scalar=(конст1 − [0x54087c]))  [1a00/1be0 = mul]
   - op(f6f8, scalar=[0x54087c]·xmm10)                     [2270/22a0]
   - bigkernel `140b60/140950(bands[i], f6f8, bands[i], nbin)`
 10. **COMBINE (52ab90–52abd7)**: acc_i = [0x5407c8+i·16];
    `op(rcx=acc_i, rdx=bands[i], r8=f6f8, r9=nbin)` [1f10→dc40 float-SUB /
    1c70→8d60 double-SUB]: **f6f8 = bands[i] − acc_i** ✓ NOTES 21b.
 11. Половины f6f8 + fma (52abd7–52ad04):
    - rbp = f6f8 + nbin (верхняя половина); op(f6f8_low?, ...) [2060/2120, scalar 0]
    - op(rcx=f6f8_upper, rdx=vec@0x5406c8, r8=acc_i) [1fa0/1940 = **3c40 fma**]
    - op(rcx=f6f8_lower, rdx=vec@0x5406e8, r8=acc_i) [1fa0/1940]
    ⇒ fma с коэф-массивами 0x5406c8 (upper/att) и 0x5406e8 (lower/rel) ✓.
 12. `acc_i += bands[i]` (52acf4–52ad03) [1b80→6840 / 1d60→**5a20** add] ✓.
 13. Ветка флага!=0 (52ad04–52ae57): повтор п.9 с теми же адресами
    (0x540698, f6f8-blend, bigkernel 1409e0/140ad0).
 14. `op(bands[i], скаляры 0x1824c4680 / xmm13)` (52ae58–52ae56 хвост).
 15. `bands[i] ⊕= track_i [rsp+0x40]` (52ae74–52ae8a) [2000/1c40].
 16. `op(vec@0x5406a8, bands[i])` (52ae8f–52aecа) [2000/1c40].
 17. флаг → bigkernel in-place 140980/1409b0 (52aecb–52af08).
 18. **БИДИР-IIR #4 ×2** (52af09–52b2af и повтор 52b0de–52b2af):
    state base 0x440510 (down@0x340510, up@0x3c0510, acc@0x440510),
    длина ctx+0x340500, in-place bands[i]. ДВА прохода подряд (каждый с reset).
 19. Финальные scale/op полосы (52b2af–52b3af):
    - скаляр `[ctx+0x1c+i·4] · xmm14` → transform [2030/1d30]
    - bigkernel in-place 1409e0/140ad0 (при флаге) ИЛИ
      op(scalar=xmm12/xmm8) [2270/22a0] + `140a40/140b00(band,band,scalar)`
 20. Инкремент i (52b3af–52b3c7), выход при i ≥ nbands.
 ### Эпилог (52b3cd–52b935)
 - GUI-snapshot блок (флаг 0x2404bc): копии векторов @0x540728 ← bands[i]
  (совпадает с consumers_out.txt:144-183).
 - Остаток до 52b935: восстановить при транскрипции (вероятно dry/wet + mirror).
 ## Состояния бидир-IIR (4 базы, лейаут {down,+0x80000 up,+0x80010/+0x100010 acc})
 | # | база | downCoef | upCoef | acc | длина |
 |---|------|----------|--------|-----|-------|
 | 1–2 (52d650+инлайн) | 0x440518 | +0x10 | +0x80010 | +0x100010 | nbin |
 | 3 | 0x2c04f8-группа | 0x2404f8 | 0x2c04f8 | 0x3404f8 | поле ctx+0x2404e8 |
 | 4 ×2 | 0x340510/0x3c0510 | 0x340510 | 0x3c0510 | 0x440510 | поле ctx+0x340500 |
 ## Открытые вопросы к транскрипции
 1. Точный op thunk 20f0/1850 в pre-combine #1 (ADD — кандидат).
 2. Аргументный порядок 2000/1c40 (axpy) и 2060/2120.
 3. Семантика bigkernel-семейства 1409xx/140axx/140bxx (какой где: exp2(-x),
   exp2(-x)·blend, mirror?).
 4. Что пишется в track_i @0x540768[i] (writer вне метода — искать отдельной охотой;
   кандидат FUN_18052e9b0).
 5. Эпилог 52b4eb–52b935.
 ## ДОПОЛНЕНИЕ 22v: декод генератора коэффициентов
 ### FUN_180530b60 (апдейт параметров, size=462)
 - fVar2 = powf([ctx+0x540878], K1); fVar3 = powf([ctx+0x54087c], K2)
  (оба live = 0.5 — параметры attack/release-класса);
 - лог-интерполяция констант → [ctx+0x540894] и [ctx+0x540898] (тау-скаляры);
 - вызовы генератора на все три состояния:
  - 180533340(ctx+0x2404e8, tau=0x540894-ветка, C=DAT_1824c459c, sr, ...)
  - 180533340(ctx+0x340500, tau=[0x540894], ...)
  - 180533340(ctx+0x440518, tau=[0x540898], ..., mult=DAT_1824c4564)
 ### FUN_180533340(state, tau, C, sr, p, mult) — ГЕНЕРАТОР КОЭФ. БИДИР-IIR
 ```
 sr' = sr · DAT_1824c3d8c
 acc = 0; downCoef[0] = 1.0 (int-пара {0, 0x3ff00000} @+0x10)
 n   = state[0];  fc_norm = (C/sr')·n
 for i in 1..n-1:
    g = (i <= fc_norm) ? fc_norm/i : powf(fc_norm/i, p)   # частотный варп!
    c  = 1/(g·tau/mult + 1)
    up[i]   = exp(c·g·tau/state[2] · DAT_1824c46b8)
    down[i] = 1 − up[i]
 ```
 ЛЕЙАУТ СОСТОЯНИЯ СОШЁССЯ: downCoef[] @base+0x10, upCoef[] @base+0x80010
 (pdVar6[0x10000] = +0x80000 байт), acc — скаляр в хвосте. Это ЧАСТОТНО-
 ЗАВИСИМОЕ сглаживание маски: сила растёт к низким бинам (1/i) с питч-законом
 выше кросса. Вот где «размазывание» нотча в реале — НЕ плоский IIR, который
 мы отвергли офлайн (22u), а пер-биновый варп!
 ### Статус сбора (решение: без коммитов до первой валидации)
 Незакоммичено: f529fe0_full.dis, BLOCKMAP_529fe0.md, thunk-резолвер,
 правки NOTES_LEVEL (22v будет добавлена при транскрипции). Следующий шаг —
 транскрипция process_frame_faithful по этой карте.
 ## ДОПОЛНЕНИЕ 23b: резолв ILT + декод FIR-цикла (52b550–52b8bb)
 ### Резолвер
 `scripts/ilt_resolve.py`: стаб = `mov eax,[rip+idx]; lea r10,[rip+tbl]; jmp [r10+rax*8]`;
 live idx=4 у всех; таблицы @0x1826xxxx. Полная таблица «стаб→impl»:
 | стаб | impl | роль (уточн.) |
 |---|---|---|
 | 0x180002210 | 0x136e0 | copy scratch→FIR? (валидатор → 39de0/3a040/3a220) |
 | 0x180002180 | 0x125e0 | complex-op A/C float (воркер 3a4a0; r9 НЕ передаётся ⇒ длина из объекта) |
 | 0x180001bb0 | 0x6a40  | complex-op A/C double |
 | 0x180001a90 | 0x5560  | complex-op B/D float — ЧИСТЫЙ FMA (fma=28, mul/add=0!) |
 | 0x1800019d0 | 0x4340  | complex-op B/D double |
 | 0x180001880 | 0x2c60  | paired-scalar op 1 (флаг-ветка) |
 | 0x180001ca0 | 0x9380  | paired-scalar op 2 |
 | 0x180001df0 | 0xb3c0  | final op float |
 | 0x180001f70 | 0xe360  | final op double |
 | 0x180140a10 | 0x141400→[1826181d8]=**0x1802a24c0** | дизайн float |
 | 0x180140a70 | 0x141580→[1826183c8]=**0x1802fa420** | дизайн double |
 | 2030/1d30 | ffe0 / 9be0 | scalar-transform |
 | 2270/22a0 | 14c40 / 15060 | scalar-transform-2 |
 | 2000/1c40 | fb60 / 8700 | axpy-класс |
 | остальное | dc40,8d60,5a20,4720,4200,3f40,5160,25e0,ee20,3c40 | как в BLOCKMAP ✓ |
 ### Дизайн-тело 0x1802a24c0 (float, вход=rcx bands[i], выход=rdx scratch@0x540628)
 Функция входа 0..0x675 (~1.6КБ), далее сиблинги. Алгоритм = **ВЕКТОРНЫЙ LOG2**
 (range reduction бит-трюком с 0.6666667, полином Хорнера на YMM-константах
 −0.5, 0.3333656, −0.2500466, 0.198225, −0.1646246, 0.1696488, −0.1517721,
 реконструкция ×ln2=0.6931472; константы @FN+0x1cdfac0..d20). Маскированный
 хвост через vmaskmovps/popcnt-таблицы (@0x181F8xxxx). ИТОГ: **scratch = log(bands[i])**.
 Диспетчер 535a70 делает swap аргументов (rcx↔rdx) перед прыжком!
 ### Структура FIR-цикла (полностью залочена)
 На полосу: 2×scalar-transform(bands[i], xmm8·[540888] при флаге 540890==0, иначе
 xmm8; затем xmm12−xmm8) [ср. decomp 184-185: A=[0x540874]−expf(K), wet=[0x540888]]
 → DESIGN=log(bands[i])→scratch → copy(scratch→FIR@0x540668) [2210]
 → opA(FIR,buf548,buf598) [2180/1bb0] → FIR[n]=0 (n=0x540534=4096!)
 → 52d920(&FIR[1], xmm13, n/2−1) деление; 52db50(&FIR[2049], xmm9, n/2−1)
 → opB [1a90/19d0] → bigkernel in-place 140b30/140aa0 (n/2+1)
 → opC [2180/1bb0] → FIR[n]=0 → 52d990(FIR, WIN_freq+n/2, n/2) УМНОЖЕНИЕ на
 **падающую половину Hann** (WIN_freq=wperiodicHann4096: w[1024]=0.5,w[2048]=1.0!)
 → 52db50(&FIR[2048], xmm9, n/2) → opD [1a90/19d0] → FIR[0]=1.0, FIR[1]=0
 → флаг-ветка: парно-скалярные 1880/ca0 → dry/wet scale ×[0x540888] над **2n флоатов**
 → final(df0/f70) с сохранённым track_i ([rsp+0x138][i]).
 ### Живые буферы (firtrace.py по s1/s2 + свежие захваты dual_b1q_0.5)
 - 0x540658 WIN_freq = периодический Hann(4096), пик 1.0 @bin2048 (НЕ рамп 0.5→0.8!)
 - 0x540758 freqaxis = линейный, шаг ≈11.713 Гц (≈48000/4096=11.71875, лёгкое
  занижение — похоже на накопительную ошибку f32 при построении суммой)
 - 0x540598 = SIMD-маски: 8×1.0 / 8×0.0 периодом 16 флоатов, ровно 1024 шт
  (YMM lane-select для комплексных кернелов)
 - 0x540550 NULL (float-путь активен, флаг 0x5408b8=0); 0x540628 scratch нулевой
  между колбеками; 0x540668 FIR = комплексная единица (1,0)×2049 бинов МЕЖДУ
  колбеками — кернел потребляется и сбрасывается внутри колбека
 - 0x5406f8 хранит аудио-масштабный сигнал (~±0.1) между колбеками — overlap/STFT
  буфер conv-движка (не «нулевой acc»!)
 - 0x540788 vs 0x5407f8 различаются (max 1.108; @171: 1.751 vs 1.409) — две РАЗНЫЕ
  кривые; ни одна не достигает нужных 3.89×@171 (11.82 дБ) ⇒ применённый фильтр
  ≠ поточечная копия любой из живых кривых (подтверждение разрыва 22y)
 ## ДОПОЛНЕНИЕ 24l: точный декод FIR-цикла по дизасму (52b550–52b8bb) + константы
 Инструмент: scripts/disasm.py (capstone, base 0x180000000 над soothe_mem.bin).
 ILT-резолв bigkernel-стабов даёт impl в 141100..141940 — все они IAT-thunk
 массивы (`mov rax,[rip+..]; jmp rax`) вне дампа ⇒ тела за импортами; НО
 последовательность и КОНСТАНТЫ цикла видны полностью:
 ### Пошагово (float-ветка, флаг 0x540890==0):
 ```
 52b5a3:  xmm6 = xmm8 · [ctx+0x540888]
 52b5bf:  th2030(bands[i], xmm6, n)          ; bands *= 1.0·s888  (xmm8=double 1.0!)
 52b5f6:  th2270(bands[i], xmm7, n)          ; scalar-transform-2
 52b62f:  535a70(scratch@540628, bands[i], n/2+1)   ; DESIGN: swap→scratch=log(bands)
 52b644:  th2210(rcx=scratch, rdx=0, r8=FIR, r9=n/2+1) ; копия (+упаковка?)
 52b696:  FIR[n]=0                            ; float-индекс n=4096
 52b69e:  52d920(&FIR[1], xmm13=-1.0, n/2-1)  ; РАЗДЕЛЕНИЕ НА −1 ⇒ ИНВЕРСИЯ ЗНАКА бинов 1..n/2!
 52b6ab:  52db50(&FIR[n/2+1], xmm9=0, n/2-1)  ; ОБНУЛЕНИЕ верхней половины до exp!
 52b6e1:  opB = th1a90(FIR, buf548, mask598, n/2-1)  ; FMA-complex (twiddle!)
 52b716:  BIGKERNEL 140b30(FIR, FIR, n/2+1)   ; EXP in-place (тело за IAT)
 52b74b:  opC = th2180(FIR, buf548, mask598, n/2-1)
 52b76d:  FIR[n]=0
 52b77c:  52d990(FIR, WINfreq+n/2, n/2)       ; окно: th2000-класс (FMA axpy!)
 52b78c:  52db50(&FIR[n/2], xmm9=0, n/2)      ; верхняя половина ×0 снова
 52b7ba:  opD = th1a90(...)
 52b7d4:  FIR[0]=1.0f; FIR[1]=0
         если флаг f890!=0:
 52b803:  th1880(FIR, {xmm10,xmm9}, n)        ; парно-скалярный с (−1|0.8, 0)
 52b81f:  th1ca0(FIR, {xmm12,xmm9}, n)        ; парно-скалярный с (1.0, 0)
 52b857:  th2030(FIR, s888, 2n)               ; ×wet (=1 live)
 52b893:  final df0(FIR, track_i, n)
 ```
 ### Константы (статические значения из дампа):
 | reg | адрес | значение | роль |
 |-----|-------|----------|------|
 | xmm8 | 1824c4140 (double) | 1.0 | множитель шага 1 = NO-OP при s888=1 |
 | xmm13 | 1824c46a0 (double) | −1.0 | делитель бинов 1..n/2 (ИНВЕРСИЯ!) |
 | xmm9 | xorps | 0.0 | обнуление верхних половин |
 | xmm10 | 1824c4680 / 1824c3e28 | −1.0 / **0.8** | парный скаляр (0.8 активен в цикле) |
 | xmm12 | 1824c3ea4 | 1.0 | парный скаляр 2 |
 | xmm14 | 1824c4674 | −0.7 | (использование вне FIR) |
 | xmm15 | 1824c4670 | −0.5 | (использование вне FIR) |
 | xmm7 | вычисл. | sens·[540870]-цепочка | аргумент th2270 |
 ### СЛЕДСТВИЯ (меняют понимание построения кернела):
 1. Перед EXP: бины 1..n/2 = −log(bands) ⇒ после exp = **1/bands** (обратная
   величина!), бины выше = exp(0)=1. Экспонента применяется НЕ к спектру маски
   напрямую — вокруг неё twiddle-FMA стадии (ops B/C/D = радиальные проходы
   FFT-класса над упакованным вещественным спектром; buf548 = cos/sin таблица,
   mask598 = lane-select).
 2. «×0.984» из 23d НЕ найден как константа цикла — либо внутри exp-IAT-обёртки,
   либо следствие нормировки twiddle-стадий. Требуется численная репликация
   пайплайна против живых захватов (rendersnap2 ph*.npz содержат готовые пары).
 3. Шаг 1 и шаг ×s888 — no-op при дефолтных параметрах (s888=1, live 24j).
 4. 52d920/52d990/52db50 = тонкие обёртки: 920→th2030/1d30 (scalar-op),
   990→th2000/1c40 (axpy!), db50→то же. «УМНОЖЕНИЕ на окно» реализовано
   axpy-кернелом, «деление» — scalar-op.
 5. Ops A–D работают над дескрипторными векторами (проверка тега [obj]==6 в
   4ca80; ошибка 0xfffffff3 при несоответствии).
 ### Динамика (попытки перехвата; окружение)
 - realtime-playback НЕ тикает DSP (нет аудио-девайса; треды спят в futex);
  кривые 22y = результат инициализации при загрузке проекта
 - `-renderproject`: полный цикл init+render+exit хоста занимает ~0.8 c
  (host@1.2s, workers@1.4s, wav@1.6s, exit@2.0s); обработка идёт в окне ~0.5 c
 - INT3-ptrace: SEIZE+TRACECLONE обязательны до CONT (иначе untraced thread
  ловит SIGTRAP и убивает процесс — подтверждено); DR-брейкпоинты: DR0@user+0x380,
  но POKEUSER DR7 даёт EIO. Скрипты: fntrace*.py, fnhw.py, fnall.py, firstop.py,
  firtrace.py, hotips.py (см. scripts/)
 - Вывод: FUN_180529fe0 и 52d650/536300/52e260 НЕ ловятся в рендер-окне —
  маск-цепь выполняется при ЗАГРУЗКЕ/изменении параметров, стационарный рендер
  использует закэшированный кернел; либо трассировать надо момент инициализации
 - Выход плагина НЕдетерминирован: md5 двух свободных рендеров различен при
  одинаковом rms (PRNG-дизеринг из LCG-прологов) — метрика только спектральная!
 ## ДОПОЛНЕНИЕ 24t: ТОЧНЫЕ СЕМАНТИКИ СКАЛЯР-ОПОВ + опознание 563a60
 Инструмент: scripts/disasm_func.py (полный дизасм функции с резолвом
 RIP-констант инлайн).
 ### Тела скаляр-трансформов (float-ветки):
 | стаб/impl | семантика | спец-случаи |
 |---|---|---|
 | th2030→ffe0 | dst[i] *= scalar | scalar==1 → скип; ==0 → zero-fill |
 | th2270→14c40 | dst[i] += scalar | scalar==0 → скип |
 Оба: скалярный хвост + AVX2 основной цикл (vmulps/vaddps ymm). Двойные
 ветки (1d30/22a0) аналогичны на sd/pd.
 ### FUN_180563a60 = init-time ПОСТРОИТЕЛЬ таблицы (не per-bin!)
 - Один caller: 1805636c2 (init-семейство).
 - Константы пролога: xmm11=8.68589 (20/ln10 — ln⇒дБ!), xmm12=1/1024,
  xmm14=0.5, xmm10=2, xmm9=−1, xmm6=1; вызов IAT 181a14cd0 (log-класс)
  над double из [obj+0x4198], ×8.68589 → дБ, запись пар во вектор
  ([obj+0xe0], рост через realloc 56c640).
 - Вывод: строит дБ-доменную таблицу парами при param-rebuild — согласуется
  с «LUT» ролью. Хвост с вирт. вызовом [rax+0x10] не декодирован.
 ### Алгебра T1: обе формы фитуют серию драйва (8 точек недоопределены)
 - softplus α·ln1p(L/β)+c: rms 0.016 (α=3.2193 β=0.4927 c=0.542)
 - LUT-форма t=((dB−A)/(B−A))^γ·M: rms 0.029 (A=−36.2 B=20.0 γ=2.097 M=10.49)
 Различить только декомпом тракта am/res→scratch или бОльшим числом точек.
 ### Открытые микровопросы
 1. Значение xmm7 на входе th2270 в FIR-цикле (52b5f6): трассировка от
   52a583; если mix^p=1 → «bands+=1» противоречит провалам ⇒ xmm7 иной
   либо порядок аргументов иной.
 2. Раскладка объекта-вектора (тег 6) для ops A–D.
 ## ДОПОЛНЕНИЕ 24u: WIN_WINDOW движка — кусочное окно аудио-пути
 Таблица WIN_WINDOW[8193] (dsp/tables_data.hpp, live-захват):
 - [0]=0.500000, плавный подъём до [2048]=0.800000, затем СКАЧОК до 1.0
  и единица до конца ([2049..8192]).
 - Длина перехода = РОВНО 2049 сэмплов = число бинов кернела (4096-сетка)!
 - Формула перехода не каноническая (ханн/синус/степенные не сошлись,
  maxdiff ≥0.09); для репликации достаточно встраивания таблицы как есть.
 - Структура намекает: блок аудио 8192 сэмплов ({16384,8192} из cfg движка),
  первые 2049 позиций получают взвешивание 0.5..0.8 (область «кернельного
  взаимодействия»?), остальное прозрачно; 0.8 подозрительно = blend.
 Проверка употребления — в resize fe00 / audio-клее (следующий раунд).
 ### Движок dc30: карта инициализаторов (24u)
 ctor 18052dc30 вызывает: 5335c0 (×2 — до/после валидатора), 534550 (×5 —
 регистрация конфиг-итемов, аргументы edx из стека [rsp+0x30..0x4c] =
 цепочка {2,4096},{16384,8192},{2,257}), затем ILT-стабы 2240/1c10.
 Следующий шаг декода: тела 534550/5335c0 + употребление WIN_WINDOW
 (по xref на таблицу или указатель из объекта).
 ## ДОПОЛНЕНИЕ 24w-3: ЖИВЫЕ КОНСТАНТЫ ГЕНЕРАТОРА 533340 (параметры частотных IIR)
 Из scalar-bank (rendersnap2 v4+) на дефолтных параметрах multi:
 ```
 p (экспонента)      = [ctx+0x54087c] = 1.000000
 tau1                = [ctx+0x540894] = 1200.000122
 tau2                = [ctx+0x540898] = 180.000015
 mult (стадия-3)     = DAT_1824c4564  = 360
 C_hz                = DAT_1824c459c  = 1000
 константы 530b60    : 0.7, 2, 800→(×2)=1600, 1200, 15, 9, 180, 0.01,
                      [540898]=180*p+xmm6 (формула видна в дизасме)
 ```
 Формула тела подтверждена дизасмом: g=min(fc_norm/i,(fc_norm/i)^p);
 c=1/(1+g*tau/mult); up=exp(...); down=1-up.
 Эксперимент: одноразовый bidir с этими коэф. НЕ двигает центр кривой
 ⇒ q-зависимость катов не через эти IIR напрямую; тракт между main-loop
 и scratch содержит ещё этапы (ops A–D / steps 9–19 BLOCKMAP).
 ## ДОПОЛНЕНИЕ 24z: FUN_180563a60 ДЕКОДИРОВАН ПОЛНОСТЬЮ — ЭТО ДЕТЕКТОРНАЯ LUT-ФОРМА!
 Постройка (на каждый банд, 1024 точки, вход double[+0x4198+i·8],
 выход пары-floats в вектор [+0xe0], stride вектора 0x18):
 ```
 dB    = log(input) · 8.68589
 idx   = i / 1024                      → store[2j]
 t     = (dB − A) / (B − A)            ; A=[sub+0], B=[sub+4] (sub=obj@[rcx+0x180])
 t     = clamp(t, 0, 1)
 если γ(=[sub+0xc]) ≠ 1:
  если [sub+0x10]==0: out = t^γ                    (классика)
  иначе             : out = sign(2t−1)·|2t−1|^γ + 1  (симметричный режим!)
 else: out = t
 out  *= 0.5                            → store[2j+1]
 ```
 Виртуальная альтернатива: [sub+0x90]!=0 → вызов [sub vtbl+0x10](A,B,dB)
 с клампом [0,1].
 ⇒ ЭТО ТА САМАЯ ФОРМА, ЧТО ДАЛА ФИТ cut=LUT(A,B,γ,M)! Стадия компрессии
 уровня найдена материально: A/B/γ живут в объекте [band+0x180],
 заполняются сеттерами. Осталось: снять их live-значения для наших конфигов
 (расширить scalar-dump на [ctx+0x180]-объект) и найти потребителя таблицы
 (интерполятор dB→out) в аудио-тракте.
 ## ДОПОЛНЕНИЕ 24cc: точная формула коэф. 533340 (дизасм+live)
 [state+8] = n = 2049 (∀ трёх состояний; live probe_states.py).
 ```
 fc_norm = (X/(sr·0.5))·n            ; X=xmm2 (C_hz=1000 или tau — уточнить)
 g(i)    = min(fc_norm/i, (fc_norm/i)^p)   ; p=[ctx+0x54087c]=1 @defaults
 c       = 1/(1 + g/mult)            ; mult=360 (стадия-3)
 arg     = |c| · g / n               ← НАСЫЩАЮЩАЯ форма (пол g→mult/n)
 up[i]   = exp(arg · (−2π))          ; down[i]=1−up[i]
 ```
 При дефолтах down ∈ [0.003..0.19] — очень мягкие сглаживатели; эффект
 только каскадом в полной цепочке шагов 9–19.
 Осталось уточнить: роль xmm13 (множитель перед c), знак/конст экспоненты,
 соответствие X=C_hz-vs-tau.
 ## ДОПОЛНЕНИЕ 24ff: семантика шага 11 (fma att/rel) — тройки (re,im,coef)
 Цепочка вызова: ee20 → 487a0 → 1065c0; воркер получает r10=n/2·12 байт
 ⇒ элементы по 12 байт = (dst_re f32, dst_im f32, coef f32).
 Call-site (52ac70–52acd4):
 ```
 шаг 11a: fma(rcx=f6f8_upper, rdx=ATT@[5406c8], r8=ACC_i, r9=n/2)
 шаг 11b: fma(rcx=f6f8_lower, rdx=REL@[5406e8], r8=ACC_i, r9=n/2)
 ```
 ⇒ семантика: f6f8[k] += coef[k]·ACC_i[k] (комплексный axpy с пер-биновым
 скаляром), верхняя половина спектра — с ATTACK коэф., нижняя — с RELEASE.
 ACC_i персистентен между кадрами ⇒ утечный интегратор: стационар
 ACC = input/(1−coef); усиление 1/(1−att): @43=2.57 @85=2.93 @171=3.42.
 Коэф. массивы СТАТИЧНЫ (∀ конфигов бит-в-бит — проверено dual/qmap/sens18).
 Гипотеза: резонансное усиление ACC объясняет k>1 аномалии уровней
 (q2: impl/ours=2.48 ≈ A(85)); точная алгебра подачи ACC в кривую — в
 остатке шагов 13–19.
 ## ДОПОЛНЕНИЕ 24gg: шаг 12 — ЭТО КОПИЯ (исправление BLOCKMAP)
 th1b80→6840→1a5a0→181a646c0 = ЧИСТЫЙ MEMCPY (vmovdqu без арифметики).
 «acc_i += bands[i]» из старого BLOCKMAP — НЕВЕРНО; реально копия
 (направление/роли rbx/rbp уточнить трассировкой регистров через цикл —
 аргументы на 52acf8: rcx=rbx, rdx=rbp, r8=n; значения rbx/rbp меняются
 через 52ac98 [0x5406e8] и др. — нужен полный dataflow-проход шагов 9–19).
 Следствие: если acc_i не накапливается сложением, «резонансная» гипотеза
 24ff требует пересмотра — возможно ACC обновляется через fma шага 11
 (ACC входит как источник), а шаг 12 синхронизирует массивы.
 ### Статус декода шагов 9–19 (24ff+24gg)
 - шаг 11: f6f8[k] += att/rel_coef[k]·ACC_i[k] ✓ (тройки re,im,coef)
 - шаг 12: COPY (не add!) ✓ исправлено
 - шаги 13–16: th2270-add ветка с xmm6=xmm12−[54087c]; bigkernel 140950;
  axpy 2000/1c40 с vec@540778 — тела не декодированы
 - полный dataflow-проход = задача следующего раунда (связная, ~сессия)
 ## ДОПОЛНЕНИЕ 24hh: DATAFLOW шагов 9–12 (точный, из дизасма)
 ```
 шаг 9a: vec698 *= (xmm12 − [54087c])          ; th1a00=MUL; при 1−1=0 → ZERO
 шаг 9b: vec6f8 += 0.8                          ; xmm10=0.8 ([1824c3e28]); th2270=ADD
 шаг 9c: bigkernel 140b60(vec6f8, bands_curve_i, vec6f8)
 шаг 10: vec6f8 = bands_curve_i − ACC_i         ; dc40: out[r8]=rdx−r10 ✓COMBINE
 шаг 11: f6f8_upper += ATT[k]·ACC_i[k]          ; тройки re/im/coef
        f6f8_lower += REL[k]·ACC_i[k]
 шаг 12: COPY(...)                              ; направление уточнить
 ```
 Регистры: rbx↔vec698/6f8/acc_i, rbp↔bands_curve/[5406e8], r14=vec6f8.
 Картина: строится КОРРЕКЦИОННАЯ кривая (bands−ACC), модулированная
 att/rel-огибающими ACC — адаптивная петля редукции.
 Остаток прохода: шаги 13–19 + эпилог + связь с FIR-секцией (52b3cd+).
 ## ДОПОЛНЕНИЕ 24ii: семантики шагов 13–16 + ИСПРАВЛЕНИЕ th2000
 th2000/fb60 (и 1c40/8700) = ПОЭЛЕМЕНТНОЕ УМНОЖЕНИЕ МАССИВОВ
 (dst[k] *= src[k]) — НЕ axpy! Исправляет интерпретации:
 ```
 шаг 13 (флаг≠0): зеркало шага 9 (vec698*=..., vec6f8+=0.8·..., bigkernel)
 шаг 14: bands_curve += (−1.0)         ; константа [1824c4680]=−1 через th2270!
        затем bigkernel 1409e0/140ad0 IN-PLACE на bands_curve
 шаг 15: bands_curve *= track_i        ; th2000 array-multiply
 шаг 16: bands_curve *= kWarp@[5406a8] ; th2000 array-multiply
 шаг 17(флаг): bigkernel in-place ещё раз
 ```
 Картинка: кривая центрируется (−1), проходит нелинейность (bigkernel,
 вероятно exp/abs — тела за IAT), модулируется track и warp.
 Против log-входа design'а значение после этих шагов должно быть >0.
 ### Bigkernel-тела за ВЛОЖЕННЫМ диспатчем
 Таблицы стабов содержат смесь IAT-слотов и внутренних адресов (напр.
 table[7]=140a00), но внутренние ведут к call runtime-helper + НОВЫЙ
 ILT-стаб с собственной idx-ячейкой (паковка/протектор). Статическое
 разворачивание обрывается. Тела bigkernel'ов (exp/abs-нелинейность шага
 14/17) остаются за пакером — при необходимости снимаются дампом памяти
 ВОКРУГ вызова в рантайме (STOP + чтение таблиц после инициализации).
 ## ДОПОЛНЕНИЕ 24jj: ТЕЛА BIGKERNEL'ОВ НАЙДЕНЫ (рантайм-резолв IAT)
 iat_name.py v2 (SIGSTOP + двойной deref + PE-экспорты) резолвит:
 ```
 стаб 140b30/140b60 → runtime 1803a06a0   (общий для float/double!)
 стаб 1409e0        → runtime 180296c80
 стаб 140ad0        → runtime 180323f20
 стаб 140a40        → runtime 1802dc0e0
 ```
 Все — НАСТОЯЩИЕ функции внутри дампа (не импорты!): большие стек-фреймы,
 x87 FNU-контроль, AVX2 полиномы, ДВЕ x87-трансцендентные инструкции
 (fyl2x/f2xm1 класс = 2^x/exp семейство). Полный декод математики каждого —
 отдельная сессия; вход/выход уже известны из контекста вызовов
 (in-place над n/2+1 элементами FIR-буфера).
 ## ДОПОЛНЕНИЕ 24jj2: R@540788 — БАНДПАС-ФОРМА, НЕ совпадающая с нашим twin!
 Сравнение (multi6, band1=500/q0.5/s12):
 ```
 бин   R@540788   наш_res    R/наш
 43    0.802      0.117      6.84
 85    1.086      0.374      2.91
 171   1.517      0.839      1.81
 342   3.521      1.401      2.51   ← ПИК ~3.5-4 кГц!
 512   2.910      1.673      1.74
 684   1.829      1.811      1.01
 1024  1.146      1.931      0.59
 1536  0.689      1.988      0.35
 ```
 Наш twin res растёт монотонно от fc; R@788 — БАНДПАС с пиком ~bin300-342
 (~3.5-4 кГц) и СПАДОМ к Найквисту. Форма напоминает кривую равной
 громкости / слухового взвешивания!
 R@5407f8 = единичная нормировка (все 1.0000 в этом прогоне).
 ### ГИПОТЕЗА (проверяемая):
 Детекторный уровень = am · ВЕС(f) / res(f), где ВЕС — кривая типа
 равной громкости (R@788?). Тест: X=am·R/res против катов multi6 —
 НЕ сошлось лобово (X@43=1.44 макс при мин кате) ⇒ взвешивание входит
 иначе (до/после res-деления, или в log-домене).
 ### Ценность
 Объясняет ВСЕ аномалии дальних бинов разом: наши дальние res слишком
 велики (нет спада), их lvl занижен, каты недобираются. Формула ВЕСА —
 ключ к кросс-конфиг параметризации.
 ## ДОПОЛНЕНИЕ 24kk3: слот [ctx+0x540668] ПОЛИМОРФНЫЙ!
 Прямой проб во время рендера: [ctx+0x540668] содержит
 `3fdbcd8940000000` = ДВА FLOAT (~0.434, 2.0), НЕ указатель!
 (пробник probe_668.py; EIO на части чтений — слот мигрирует).
 ### Следствия:
 1. rendersnap2 пропускает слот 668 (ptr<0x10000 или мусорный ptr) ⇒
   во ВСЕХ новых захватах (sc_multi4b, sc_tt*, sc_q*, ...) НЕТ FIR-массивов.
 2. Старые захваты rendersnap.py v1 ИМЕЛИ валидный FIR-указатель в эти
   моменты (ловили фазу обработки). Данные старых phase*.npz про FIR —
   валидны для своих моментов, но смешивать с новыми нельзя.
 3. Все «FIR mag» анализы через этот слот зависят от ТОГО, в какой фазе
   слот был пойман: указатель-на-буфер vs скаляры vs сброс.
 4. Значения скаляров (~0.43, 2.0) — кандидаты: att/rel? dry/wet? g-компоненты?
 ### Статус
 Канонический путь чтения ПРИМЕНЁННОЙ маски: слот [ctx+0x540678] (кривая
 банды) — он стабилен и МАТЧИТ АУДИО в deep-фазах (±5%).
 ## ДОПОЛНЕНИЕ 24mm2: ПОЛНАЯ РЕЗОЛЮВСЯ ВСЕХ 10 ЯДЕР + ИСПРАВЛЕНИЯ DATAFLOW
 Инструмент: статический резолв цепочки стаб→таблица→L2→IAT-слот по
 soothe_mem.bin (без live). Стаб = `movsxd rax,[idx@1826159a0]; lea r10,[tbl];
 jmp [r10+rax*8]`, idx=4, L2=`mov rax,[slot]; jmp rax`.
 ### Таблица резолва (исправляет 24jj!)
 | стаб | таблица | runtime | опознание по константам |
 |------|---------|---------|------------------------|
 | 140950 | 182617448 | **18026b820** | exp2/exp DOUBLE (ln2, log2e, 1021.5, 2^27) |
 | 140980 | 182617488 | **18027c120** | **logf** FLOAT (ряд −½,+⅓,−¼,+⅕,−⅙; ln2 hi/lo; 2^32) |
 | 1409b0 | 1826174c8 | **18028d1e0** | **powf/log+exp** DOUBLE (ряд log + магия expf вместе) |
 | 1409e0 | 182617508 | 180296c80 ✓ | **expf** FLOAT — ДЕКОДИРОВАН ПОЛНОСТЬЮ (ниже) |
 | 140a40 | 182617588 | 1802dc0e0 ✓ | exp-вариант FLOAT c hi/lo сплитами |
 | 140aa0 | 182617608 | **18030fee0** | **sincos** DOUBLE (1/6,1/120,1/5040; π hi/lo) |
 | 140ad0 | 182617648 | 180323f20 ✓ | кусочно-табличная DOUBLE (сетка Δ=0.00541521) |
 | 140b00 | 182617688 | **180367980** | pow/exp DOUBLE (1023/1022, магия 1.5·2^20) |
 | 140b30 | 1826176c8 | **1803831c0** | кусочно-табличная FLOAT (π/2, π/4, сетка 184.665!) |
 | 140b60 | 182617708 | 1803a06a0 ✓ | **DIVIDE** FLOAT B/A (rcp+квантование+vpermps-таблицы+полином невязки) |
 **ИСПРАВЛЕНИЕ 24jj**: «140b30/140b60 → общий 1803a06a0» — НЕВЕРНО.
 FIR-секция вызывает 140b30 = 1803831c0 (табличная кривая), divide только в шаге 9c.
 ### ДИСПЕТЧЕР float/double
 Каждый call-site имеет ПАРУ стабов через `call [181bab008]; test eax,eax; jne`:
 float-стаб (eax==0) / double-стаб. Дескрипторы type_info СТАТИЧЕСКИ идентичны
 (оба →182650db8) ⇒ eax=0 ⇒ **double-ядра мертвы на нашем пути**; рендеры
 идут по float. Double-тела не транскрибируем (отмечено на будущее M8).
 ### expf 180296c80 — полная формула (горячий цикл, FMA-точно)
 ```
 n   = fma(log2e_hi=1.4427, x, MAGIC=12582912.0)     ; округление до int
 k   = n − MAGIC
 r   = (x − 0.693146·k) − 1.42861e-06·k               ; ln2 hi/lo
 p   = (((0.00829172·r + 0.0418735)·r + 0.166674)·r + 0.499994)·r + 1)·r + 1
 out = bits( (k<<23) + bits(p) )                      ; vpaddd сборка
 guard: |x|>87.3365 → slow-path; head/tail через vmaskmovps+popcnt-маски
 ```
 Коэф. минимаксные — транскрибировать КАК ЕСТЬ.
 ### DIVIDE 1803a06a0 — структура (90%)
 ```
 A=[rcx], B=[rdx], dst=[r8]; r9d=n
 q0 = rcp(A); q0 += 2^23-magic (округление); q = q0 & 0xfff00000  ; 12 бит
 e  = (q>>23); idx = q>>20 → vpermps tbl@1821269c0 (127±ε) и @182126a00
 err = 1 − q·A
 полином невязки {0.207515, −0.241687, 0.288535, −0.360671, ..., 0.240264, 0.0555119}
 сборка через магию 1.5·2^20 + vpslld 20
 результат ≈ B/A с точностью ~0.5 ulp
 ```
 Таблицы коррекций сдамплены (per-mantissa-top-bits).
 ### ИСПРАВЛЕНИЯ DATAFLOW (по fn529fe0.dis, адреса call-sites)
 1. **Шаг 14 порядок ОБРАТЕН к BLOCKMAP 24ii**: сначала `bigkernel exp IN-PLACE
   на bands_curve` (52ae0e), ПОТОМ `bands_curve += (−1.0)` (52ae40, конст.
   1824c4680 через th2270).
 2. Шаг 9b точно: `vec6f8 += [ctx+54087c] · 0.8` (xmm10=0.8@1824c3e28,
   множитель виден в asm: mulss xmm6,xmm10 после movss xmm6,[54087c]).
 3. Шаг 9a: `vec698 *= (xmm12=1.0 − [54087c])` ⇒ zero-fill при дефолтах ✓.
 4. Шаг 10 combine dc40: аргументы rcx=ACC_i(**таблица указателей @0x5407c8**,
   НЕ дампилась rendersnap2!), rdx=bands_curve_i(@678+i), r8=vec6f8(@6f8),
   семантика dst=r8: vec6f8 = bands_curve_i − ACC_i. ACC-слот надо ДОБАВИТЬ
   в SLOTS rendersnap2 (0x5407c8).
 5. Шаги 15/16 подтверждены: th2000/th1c40 array-mul; затем rbx=[5406a8]
   (kWarp) — array-mul на bands_curve.
 6. Эпилог: скалярная часть из decomp (consumers_out 100-143): mix-веса,
   `fVar17 = [540874] − expf(DAT_1824c4704=-ln1000)` → bands += f17·[540888].
 ### Call-site карта больших ядер (fn529fe0)
 ```
 52a63a/52a641: 140980(logf-float)/1409b0  — pre-combine #1
 52ab84/52ab8b: 140b60(divide)/140950      — шаг 9c
 52acd?       : (шаги 10–12 мелкие ILT)
 52ae0e/52ae15: 1409e0(expf)/140ad0        — шаг 14 нелинейность
 52b32c/52b336: 1409e0(expf)/140ad0        — шаг 17 (повтор)
 52b3a0/52b3aa: 140a40(exp-var)/140b00     — пост-17
 52b716/52b71d: 140b30(кривая-float)/140aa0 — FIR-секция
 ```
 ## ДОПОЛНЕНИЕ 24mm5: ПОЛНАЯ КАРТА ТРАКТА — буферы каждого шага; design = conv-тело 22z
 ### Полоса-цикл (float-путь), трасса регистров 52a580–52b3cd
 ```
 пре:    [678i] *= скаляры (s888-цепь, xmm7·[540870]·[54088c])
        LOG#1 (140980!) на [678i]           ; 52a63a — В ЛОГ-ДОМЕН заранее
        combine 52d650([678i],[6f8])
 шаг 9a: vec698@698 *= (1−[54087c])          ; zero
 шаг 9b: vec6f8@6f8 += [54087c]·0.8
 шаг 9c: DIVIDE dst=[678i]: A=arg(rcx)=[678i], B=arg(rdx)=[6f8]
        ⇒ [678i] = vec6f8 / bands_curve     ; in-place
 шаг 10: dc40: rcx=ACC_i(@7c8+i!), rdx=[678i], r8=[6f8]
        ⇒ vec6f8 = bands_curve − ACC_i      ; ACC — таблица указателей 7c8
 шаг 11: fma ATT(@6c8)/REL(@6e8) — пары вызовов 1fa0/1940
 шаг 12: COPY 1b80/1d60 c [678i]
 шаг 13: зеркало 9a/9b + оп 1eb0(cbe0)([678i],[6f8])
 шаг 14: EXP#1 (1409e0=expf) на [678i]; затем += (−1)
 шаг 15: array-mul: X[rsp+0x40] *= [678i]    ; НЕ bands*=track!
 шаг 16: [678i] *= kWarp@[5406a8]
        LOG#2 (140980) на [678i]            ; 52aefd — возврат в лог!
        IIR4 ×2 бидир                       ; ~52af09–52b2b6, СПЕКТРАЛЬНОЕ
                                            ; СМЕШЕНИЕ В ЛОГ-ДОМЕНЕ
        скаляры xmm14(−0.7)/xmm15(−0.5)-класс
 шаг 17: EXP#2 (1409e0) на [678i]; += scalar; exp-var 140a40 финал
        → bands_final @678i
 ```
 ### FIR-секция (52b3cd–52b94a)
 ```
 bands_final *= s888, *= [540888]; += xmm7 (скаляр с expf(−ln1000)=0.001)
 DESIGN: call 535a70(rcx=scratch@628, rdx=bands)
  535a70 = диспетчер СО СВОПОМ аргументов → ILT 140a10/140a70 →
  → РЕЗОЛВ: float=1802a24c0 (!!!), double=1802fa420
  ⚡ ЭТО ТЕЛО FFT-CONV ИЗ ОТКРЫТОГО ВОПРОСА 22z («conv_float_a24c0.dis»,
    184K AVX2). Дизайн детектора == недекодированный conv. Пазл склеен.
 дальше: copy th2210; complex-op th2180/th1bb0 с твидл-буферами
        548/550/598; знак −1 (52d920); EXP 140b30(=1803831c0);
        окно 52d990(WINfreq); pair-scalar 1880/1ca0; *= wet[540888];
        финал df0(FIR, track_i)
 ```
 ### Где γ=1.760561
 mask = bands_final^γ точно ⇒ γ возникает между scratch=log(bands_final)
 и финальной маской: либо ВНУТРИ design 1802a24c0 (масштаб на выходе),
 либо в комплекс-op цепочке 52b644–52b716 перед EXP 140b30. Обе точки
 локализованы до ~десятка инструкций — декод следующего раунда.
 ### Исправление понимания слотов
 - 688 = exp(628) тривиально: 628 — копия лога bands_final (design),
  688 — сами bands_final (или их exp-копия). «track» — имя рендерснапа.
 - 678 ПОСЛЕ цикла = bands_final; применённая маска перезаписывает
  поверх (финальный combine) — поэтому захваченный 678 матчит аудио.
 ## ДОПОЛНЕНИЕ 24mm6: ПЕРЕД EXP В FIR — УМНОЖЕНИЕ НА 2.0 (не −1!); гипотеза γ=2·k_design
 ### Точная последовательность 52b60c–52b720 (проверено, без пропусков)
 ```
 rcx=[540628](scratch), rdx=[r15](источник design — уточнить r15!)
 call 535a70 → swap → 1802a24c0(scratch ← DESIGN(src))
 th2210: FIR(@540668) ← scratch (copy, edx=0)
 opB: th2180(FIR, buf548|550, buf598)          ; complex pass
 FIR[n]=0
 FIR[1 .. n/2] *= xmm13 = 2.0 @1824c41e0       ; 52d920, БЫЛО «−1» в 24l — НЕВЕРНО
 FIR[n/2+1 .. n-1] *= xmm9 (=0)                ; 52db50
 opC: th1a90(FIR, buf548|550, buf598)          ; complex pass
 EXP in-place 140b30 (float) / 140aa0 (double)
 ```
 xmm13/xmm9 не перезаписываются между 52b3d6 и использованием (проверено).
 ### Гипотеза источника γ
 Если opB/opC сохраняют пропорциональность (упаковка real-FFT), то
 mask = exp(2 · scratch) ⇒ γ = 2·k, где k — масштаб выхода design
 1802a24c0 относительно ln(bands): k = 1.760561/2 = 0.8802805.
 Альтернатива: k=1, а opB/opC суммарно дают множитель 0.88028.
 ### Открытые микровопросы (следующий раунд, всё локализовано)
 1. Что такое [r15] на входе design (bands_final@678 или иной буфер)?
 2. Семантика opB/opC (th2180/th1bb0/th1a90/th19d0 + твидлы 548/550/598)
   — вероятно упаковка/развёртка real-FFT.
 3. Масштаб выхода design: декод хвоста 1802a24c0 (файл уже есть:
   nls_dasm/conv_float_a24c0.dis, 184K).
 4. Согласование с identity-фазой захватов (гонка финального combine).
 ## ДОПОЛНЕНИЕ 24mm7: design выход = точный ln(bands_final); γ создаётся после design
 ### Численный тест (multi6/ph034, identity-фаза)
 ```
 scr@628 − ln(cur@678): max|r| = 9.0e-08 (float32 eps) на 1013 бинах
 ⇒ k_design = 1 (в момент захвата)
 ```
 Оговорка: станционарность делает «свежий» и «сталый» scratch
 неразличимы; но факт (scr, cur)=(лог, значение) одной пары твёрд.
 ### Следствие для γ
 γ=1.760561 ≠ 2 ⇒ множитель НЕ только «×2 перед EXP». Источники:
 (a) opB/opC не взаимно сокращаются (не чистая упаковка real-FFT);
 (b) пост-exp шаги: окно 52d990 (варьируется по позиции — нарушил бы
    степенной закон, значит действует на верхнюю половину/после),
    pair-scalar th1880/th1ca0, финальный combine df0(FIR, track_i),
    где track=exp(scr)=bands_final.
 Комбинации дающие γ из {1,2}: 1+2x=1.760561 ⇒ x=0.3802805;
 либо лог-доменное смешение track^a·FIR^b c a+2b=1.760561.
 ### Статус декода design 1802a24c0
 AVX-512 (zmm, masked {k3}/{k4}), 3822 строки objdump — трансформ-класс.
 Для замыкания γ его полный декод МОЖНО НЕ НУЖЕН: достаточно семантики
 opB/opC + df0 (десятки инструкций в fn529fe0.dis).
 ## ДОПОЛНЕНИЕ 24mm8 (финал захода): opB/opC/df0 резолвлены
 ```
 opB: 180002180→180004ca80(f)/18001d160(d)   ; дескриптор-оп (тег [obj]==6)
 opC: 180001a90→18001a0c0(f)/180018400(d)
 df0: 18000df0→18000b3c0 ; f70→18000e360     ; финальный combine
 ```
 Все четыре микровопроса 24mm6 закрыты или локализованы до тел-обёрток.
 Следующий раунд: семантика 4ca80/1a0c0 (кандидаты источника γ=2k−масштаба),
 затем полный numpy-конвейер.
 ## ДОПОЛНЕНИЕ 24mm9: opB/opC = RFFT-близнецы; df0 = complex-mul; цепь валидирована 0.0065 дБ
 ### Слой вызовов FIR-секции (уточнение поверх 24l/24mm6)
 ```
 обёртки: th2180 impl=125e0, th1a90 impl=5560 — только перестановка аргументов:
         воркер получает (rcx=data, rdx=data, r8=ПЛАН, r9=WORK), ин-плейс.
 ПЛАН = [ctx+540548] (buf548!): tag=6 [+0], log2n=12 [+4], flag [+8]=0,
       scale_flag=1 [+0xc], scale=2^-12 [+0x10], workbytes=16384 [+0x18].
 WORK = [ctx+540598] — рабочая область FFT (заметение «lane-mask» из 23b).
 th2180 → воркер 4ca80(f)/1d160(d): INVERSE real-RFFT (голова: X[0]±X[Nyq]).
 th1a90 → воркер 1a0c0(f)/18400(d): FORWARD real-RFFT (хвост: пакинг Nyq).
 тела: импортные близнецы 181b853e0(inv)/181b81b80(fwd); константы только
      ±0.707107; масштабов нет. ffe0 = ×scale pass (skip при scale∈{0,1}).
 copy th2210 → 136e0 → 4d900(src,dst,n): pack re=v, im=0 (vunpcklps+zero).
 df0 18000b3c0: ПОЭЛЕМЕНТНОЕ КОМПЛЕКСНОЕ УМНОЖЕНИЕ dst=[rdx]=arg2:
      track_i := track_i ⊗ FIR (vfmaddsub213ps; f70/b560 — double версия).
 EXP 140b30 → 1803831c0: полиномиальная комплексная exp (без таблиц значений):
      magic 12582912 (=2^23·1.5), guard 87.33654, редукция 184.665≈128/ln2,
      коэф. {−0.01604,−1.541667(=−37/24), 3.166e-05, 1.008329, −1.65777e-06,
      −0.01932, 0.00134, 0.00541687, 10000, 4.19179}; AVX-512+FMA.
      Численно = поточечный комплексный exp (flat-exp проигрывает 8 дБ).
 ```
 ### Полная последовательность (52b60c–52b893, все шаги, без пропусков)
 ```
 design 535a70(scratch@628 ← ln(bands_i))          ; 52b62f, своп аргументов
 copy 2210(scratch → FIR, 2049 пар (re,im=0))      ; 52b644
 FIR[4096]=0                                       ; 52b685 Найквост ДО фолда
 inv-RFFT opA                                      ; 52b672 th2180
 fold: float[1..2047]*=2.0 (xmm13@1824c41e0)       ; 52d920
      float[2049..4095]=0                         ; 52db50
 fwd-RFFT opB                                      ; 52b6e1 th1a90
 EXP in-place, аргумент×q (q≈0.80, источник ОТКРЫТ); 52b716
 inv-RFFT opC                                      ; 52b74b th2180
 FIR[4096]=0                                       ; 52b76d
 float[0..2047]*=WINfreq[2048..4095]               ; 52d990 (падающий Hann)
 float[2048..4095]=0                               ; 52db50
 fwd-RFFT opD                                      ; 52b7ba th1a90
 FIR[0]=1.0f; FIR[1]=0                             ; 52b7cd
 (flag f890≠0: pair-scalars 1880/ca0 — live мертво)
 th2030(FIR, wet=s888, 2n float)                   ; 52b857, s888=1 no-op
 df0(FIR, track_i, n): track_i := track_i ⊗ FIR    ; 52b893
 ```
 Смысл: классическое минимально-фазовое ядро через кепстр
 (IDFT лога → фолдинг ×2 причинной части + усечение → exp → обратный ход).
 ### Валидация и γ
 mask_sim = trk·|F(q)|: 60 ультрачистых кадров, ВСЕ 2049 бина:
 rms мед 0.0065 дБ / p90 0.0075 / max 0.035 при q=0.80 (порог 0.05 ✓).
 γ = 1 + s_F(q), s_F = наклон log|F| по log trk в нотче: q=0.8 ⇒ γ_pred=1.7516
 (точный 1.760561). Открыто: место q в асме (внутренность 1803831c0);
 unicorn не эмулирует FMA ⇒ нужен статдекод ядра или live-захват входа EXP.
 Дизасмы: /tmp/opencode/cascade/{wrapA_125e0,wrapB_5560,opB_4ca80,opC_1a0c0,
 df0_b3c0,h_ffe0,h_136e0*,imp_b8*3e0_full,bk_1803831c0}.dis
 (*copy: python3 scripts/disasm_func.py 1800136e0 — ВАЖНО: полный VA,
 короткая форма «125e0» даёт пустой файл!).
@@ -0,0 +1,74 @@
 # PROMPT FOR NEXT SESSION (2026-08-25, после 24kk2)
 Продолжаем bit-exact реверс soothe2 в /home/m/re-tools (ветка main).
 ГЕЙТ СМЕНЫ КАНОНА = BIT EXACT (решение пользователя): все параметры
 прослежены до декомпа + корпус в шумовом пол. До тех пор канон не трогаем.
 ПРОЧИТАТЬ ПЕРВЫМ: AGENTS.md (фаза-заголовок 24kk2 + env-флаги + инструменты)
 → handoff/NOTES_LEVEL.md обновления 24j–24kk2 → BLOCKMAP_529fe0.md
 (дополнения 23b/24l/24hh/24ii).
 ## СОСТОЯНИЕ
 ПРИМЕНЕНИЕ ДЕКОДИРОВАНО ДО ФОРМУЛ:
 ```
 mask(b)  = 10^(−cut_D(b)/20)                вещественная, per-bin multiply кадра
 cut_D(b) = α·ln1p(lvl_raw(b)/β)+c  [+Δ вторые пики]
 lvl_raw  = am/res·scale                     наш фронтенд, float-parity ✓
 слой     = STFT БЕЗ синтез-окна             RT_SYN=1
 ```
 Калибровки (rms ≤0.016 дБ):
 ```
 dual fc500 q0.5 s12 ДВА тона : α=3.2193 β=0.4927 c=+0.54
 fc1000 q0.5 s12 ОДИН тон     : α=1.6151 β=0.3645 c=+0.48
 fc500 q0.5 s12 ОДИН тон      : α=1.1530 β=0.4038 c=+0.33
 ```
 Корпус: dual **0.193 max 0.438** с флагами `RT_VLAW=1 RT_SYN=1 RT_NOWARP=1
 RT_NOIIR3=1 RT_IIR12=0`. Канон TOTAL 2.286 нетронут.
 ОТЗЫВАНО И НЕ ВОЗВРАЩАТЬСЯ: ×1.805-свёртка [23e], OLA-нормировка [24i],
 двухстадийный γ₀ как множитель закона [24m — это артефакт двух тонов],
 клампы параметров [24dd], B∝am [24bb], axpy-семантика th2000 [24ii — это
 array-multiply], «acc_i += bands[i]» шага 12 [24gg — это COPY].
 ## ЗАДАЧА №1: каскадный симулятор шагов 9–19 (оп-за-опом)
 Dataflow декодирован (BLOCKMAP 24hh/24ii):
 ```
 vec698 *= (xmm12−[54087c])            ; обнуление при дефолтах
 vec6f8 += 0.8                          ; базовая линия
 bigkernel(vec6f8, bands_i, vec6f8)
 vec6f8 = bands_curve − ACC_i           ; COMBINE (dc40: out=rdx−r10)
 f6f8_верх += ATT[k]·ACC[k]; низ += REL[k]·ACC[k]   ; тройки re/im/coef
 COPY(...)                              ; шаг 12 = memcpy!
 bands += (−1.0); bigkernel in-place    ; центрирование+нелинейность
 bands *= track_i; bands *= kWarp       ; th2000 = array-multiply
 IIR4 ×2; финальные scale/op
 → bands[i] → design log → scratch → FIR → audio multiply exp(scratch)
 ```
 Тела bigkernel'ов по рантайм-адресам: 1803a06a0/180296c80/180323f20/1802dc0e0
 (x87 exp-семейство). Коэффициенты IIR-генератора 533340 исправлены
 (насыщающая форма arg=|c|·g/n; p=[54087c]=1, mult=360, C=1000).
 Метод проверки: собрать симулятор в Python (numpy), прогнать lvl_raw из
 tract_* через каскад, сравнить с deepest-scratch кривыми sc_* — rms < 0.05 дБ
 = замкнулось. Затем перенос в C++.
 ## ЗАДАЧА №2: параметризация α(контент)/fc
 α удваивается с числом тонов (частотное смешение шаблонно-локальное —
 далёкий тон не влияет, 24ll). Инструмент: scripts/campaign.py (ячейка ≈8 мин).
 Двухтональная дистанционная серия уже снята (sc_d*, инверсия не сошлась —
 нужен каскад из Задачи №1 сначала!).
 ## СРЕДА (грабли, всё проверено болью)
 - rendersnap2.py: RENDER_FILE брать ИЗ rpp (клоны наследуют путь — однажды
  перезаписали реф); каталоги снапов задавать уникальные (argv[3]).
 - touch+mtime перед каждым замером (same-second cmake hazard).
 - tone1k.wav громче dual.wav в 2.28× — при сравнении серий учитывать.
 - Свип НЕ годится для Y/X-отношения; только мультитон/мультиуровень.
 - Динамика V(t)↔g(t) на рампе не сходится — пары только из стационара.
 - HW-ловушки под wine невозможны (wine держит слоты); INT3 требует
  дисциплины fnwatch4 и ГАРАНТИРОВАННО свежего рендера.
 - Метрика только честная: render_parity.load / Гёрцель последних 0.75 c.
 - Коммитить подшагами; факты → NOTES_LEVEL (очередной номер 24ll+).
@@ -0,0 +1,44 @@
 # Runtime capture — live DSP tables (registry heartbeat, SR 48000)
 > **Сжато 2026-08-28.** Полный журнал 2026-08-19—2026-08-23 → [`handoff/archive/NOTES_LEVEL_2026-08-18_2026-08-23.md`](archive/NOTES_LEVEL_2026-08-18_2026-08-23.md) + [`handoff/NOTES_LEVEL_INDEX.md`](NOTES_LEVEL_INDEX.md). Детали FIR-цепи → `handoff/BLOCKMAP_529fe0.md`.
 ## Registry heartbeat
 Heap run `{u64 count, u64 ptr}` stride 0x10, base `0x29b06c0` (anon 0x2922000) — офлайн, `0x28b06c0` — GUI (сдвиг ареной). Поиск: скан `u64==8193` + readable ptr, требовать run `count/ptr` stride 0x10. Надёжный beacon (vptr/44100-marker — тупик).
 ## Captured tables (SR=44100 project → internal freq-axis 48000)
 | idx | count | content |
 |-----|-------|---------|
 | [00] | 8193 | identity ~1.0 |
 | [01] | 8193 | **WIN_freq 0.5→1.0** (idx1024 0.68, idx2048 0.8) = live `0x540658` |
 | [03] | 8193 | **0.596→0.126** = `rwin_C0` (byte-identical GUI) |
 | [04] | 8193 | 0.404→0.874 = 1−[03] |
 | [05] | 8193 | 0.0435→~0 |
 | [06] | 8193 | 0.9565→~1.0 |
 | [0d] | 2049 | **freq-axis 0..23988.3 Hz, 11.713 = 48000/4096** → internal SR 48000 |
 | others | 8193/16384 | опоры, FFT-планы |
 **Проверка:** `[03]` head 0.5960761 byte-identical GUI — registry authoritative. **Freq-axis spacing 11.713 ⇒ internal SR 48000** независимо от проекта 44100.
 ## Files saved
 - `handoff/rtwin_freq_44100.npy` — WIN_freq 8193 f32 (live 0x540658)
 - `handoff/rtfreqaxis_48000_internal.npy` — 2048 f32
 - `handoff/rtwa_596.npy`, `rtwb_404.npy`, `rtwc_043.npy`, `rtwd_956.npy` — [03]–[06]
 - Raw snapshot: `/tmp/snap_all.bin` (318MB, chunked 8MB pread — иначе EIO)
 ## Method (repro)
 - `rtsnap_fast.py`: spawn `reaper -renderproject render_long`, find host `soothe2` in maps, sleep 6s, chunked pread ALL → `snap_all.bin`. Scan 0.1s/318MB.
 - `rtsnap2.py`: two-point diff — registry byte-identical (stable).
 - Live ctx `0x2370040` (`+0x24==48000`, `sens>100`) — единственный populated, остальные empty. `handoff/rtctx_live.json`.
 ## Live ctx (realtime playback, `rtctx_live.json`)
 - Pointers `0x540688` identity, `0x540698` window, `0x5406b8/c8/d8/e8` WA/B/C/D, `0x540748` warp, `0x540758` freqaxis (`v85=995.6 Hz`).
 - Scalars: `0x540870=440.955`, `0x540880=25.0`, `0x540884=10.0`, `0x540888/88c=1.0`, `0x54087c=1.0/10.0` (selectivity/sharpness).
 - `acc/f6f8` arrays **zero** в стационаре playback — combine idle.
 - Bands curve `R=1/mask` peak следует fc (bin43@500 → bin85@1000), `0x5407f8` min 1.0. Применённый фильтр ≠ pointwise копия R (нужен FFT-conv).
 Дальше — `handoff/NOTES_LEVEL.md:24mm11+` (live ptrace) + `handoff/BLOCKMAP_529fe0.md`.
@@ -0,0 +1,46 @@
 # NOTES_LEVEL — оглавление журнала
 > **Навигатор:** `NOTES_LEVEL.md` — живая голова (25a+, 2026-09-02, ~470 строк). Архив 2026-08-18—2026-08-23 → [`handoff/archive/NOTES_LEVEL_2026-08-18_2026-08-23.md`](archive/NOTES_LEVEL_2026-08-18_2026-08-23.md) (4024 строки). Статус TOTAL → [`README.md:13`](../README.md).
 ## Живая голова (`handoff/NOTES_LEVEL.md`)
 | Дата | Заголовок | Суть |
 |------|-----------|------|
 | 24mm5–24mm7 | Полная буферная карта, DESIGN 1802a24c0 | Два круга log→exp, bidir×2 в лог-домене |
 | 24mm8-бис | Точка входа следующего раунда | opB 18004ca80, поиск γ |
 | 24mm9 | FIR-цепь MIN-PHASE, 0.0065 dB | Цепь 52b60c+, RFFT близнецы, df0 complex-mul, валидация |
 | 24mm10 | EXP ядро полностью, парадокс q | Формула 1803831c0, нормировки s_f/s_i=1 |
 | 24mm10-бис | Итог близнецов radix-4 | INV/FWD 2048, твидлы 548, без масштабов |
 | 24mm11 | LIVE ptrace, цепь до df0 бит-точна | wine_ptrace_trace, Y_model=1.0, df0 complex-mul |
 | 24mm12 | Детекторный каскад vt+0x28=180529c60 | Оркестратор 5300f0, vtable карта, 0x281 байт |
 | 24mm13 (+доп) | Хелперы каскада 529c60 | 5355d0→16140, 530080, 20f0/1850/1a00, рекуррентия Haar |
 | 24mm14 | Каскад декодирован — 3 фазы | `|z|` → Haar×2 → peak/sin/w/blend 5407a8 |
 | 25a (2026-09-02) | chain_9_19 staged, IIR4-генератор, k-mapping dedup, аудит | IIR4 double (movsd), generate_iir4_coefs, RT_CASC/RT_IIR4_GEN gates, baseline 2.689 |
 ## Архив (`handoff/archive/NOTES_LEVEL_2026-08-18_2026-08-23.md`) — по периодам
 | Период | Ключевые темы | Где |
 |--------|---------------|-----|
 | 2026-08-18 | Две twin-цепочки, FUN_180563440/563ce0, 530d30 веса, FUN_180529fe0 consumer, 0x530d30 NEGATIVE, bridge B.12 | строки 1–300 |
 | 2026-08-18h2–h3 | Runtime capture, поле 0x540658, registry 0x28b06c0/0x29b06c0, window 0.5→0.8, freq-axis 48000 | 475–580 |
 | 2026-08-19 | LUT-leg joint fit, level-tracker, decomp lock, registry heartbeat, P4 mask-chain, SR mismatch | 580–900 |
 | 2026-08-20i–x | P2 kernels 8d60/3c40, twin-mask, F0 gate determinism, parametric LUT benches, wrappers | 900–1100 |
 | 2026-08-21b–d | IIR3 bidir, combine scope, structural corpus 2.286, reduction-law affine, Phase A | 1100–1360 |
 | 2026-08-22a–f | Phase B refuted, Step7 GUI-only, param bridge, floor −20.72 blend·ln10/20 | 1360–1600 |
 | 2026-08-22g–p | NO-LUT, white-noise probe, bare chain, pool/scale/floor, joint-fit, dual diag, affine port | 1600–1970 |
 | 2026-08-23a–z | Скалярный закон насыщен, 529fe0 по raw asm, faithful-цепь, теорема α, спрединг отвергнут | 1970–2300 |
 | 24j–24ab | Применение декодировано: A=1.019·V^1.8345, FIR=exp(scratch), двухстадийность γ₀=1.79 | 2750–3550 |
 | 24bb–24kk3 | k-маппинг, q-независимость, далёкий тон локально, дистанционная серия, симулятор не замкнулся | 3550–4000 |
 | 24ll–24mm4 | multi6, g_k пер-пиковый, archive закрытие, γ=1.760561, точ. законы отвергнуты | 4000–4024 |
 ## Как читать
 1. **Старт:** `README.md:13` (статус) → `AGENTS.md` (runbook) → `BITEXACT_PLAN.md:1` (3 шага) → этот индекс → `handoff/BLOCKMAP_529fe0.md`.
 2. **Детали по теме:** `rg -n "24mm9\|Q-НЕЗАВИСИМОСТЬ\|Haar"` в соответствующем файле.
 3. **Полный поиск:** `rg -n "cut_D\|lvl_raw\|5407c8" handoff/archive/NOTES_LEVEL_2026-08-18_2026-08-23.md handoff/NOTES_LEVEL.md`.
 ## Дубли устранены
 - `UPDATE 2026-08-22g` дубль (строки 1577/1591 в архиве) — логически один.
 - `UPDATE 2026-08-20s` дубль (974/977) — один.
 - Заголовки унифицированы: `## ============ UPDATE` → `##` в живой голове; архив сохранён как есть для истории.
@@ -1,76 +1,29 @@
-# Twin FUN_180535880 transcription (Phase 2)
+# Twin FUN_180535880 — краткая справка (детали → `dsp/twin.cpp`)
-## PHASE-3 CALLER + GRID (2026-08-18)
+> **Статус:** транскрибирован, gate PASS `dsp/build/twin_check` max rel err 1.27e-5 (float-parity). Полный журнал 2026-08-18 → [`handoff/archive/NOTES_LEVEL_2026-08-18_2026-08-23.md`](archive/NOTES_LEVEL_2026-08-18_2026-08-23.md) + [`handoff/NOTES_LEVEL_INDEX.md`](NOTES_LEVEL_INDEX.md).
 - Caller FUN_180536300 (decomp 6218 / sp_180536300.txt) decoded at buffer level:
  - params: param_1=top DSP, param_2=out, param_3=?, param_4=band obj, param_5=cell count,
    param_7=scratch (>=52*N bytes complex).
  - scratch: +0..8N A-accum, +8N..16N B-accum, +16N..24N z-powers, +24N..32N rotor-in
    (fill 1.0), +32N..36N phase ramp lVar1, +36N..44N rotor-out z1 (r14), +44N..52N
    lVar3(1.0)/lVar4(0.0) for interleave+negate+exp build.
  - 0x18052da00(lVar1, scale=2pi/(iVar7*sr), N) = ramp fill (float, via 0x180001a00);
  - twin always = FUN_180535880 (type-info compare of same addr -> equal);
    sibling FUN_180536f90 is dead.
  - 0x18052dbc0(param_2, lVar4, N) = cplx interleave -> out=(cos,sin)=e^{+i theta}.
 - GRID LAW (verified from trace.txt case8 + constructor):
  - os = *(int*)(param_1+0x240080) = 4   (written 0x18052ce3e `mov ...,0x4`)
  - sr = *(float*)(param_1+0x24) = 44100.0
  - fs_total (generator) = os*sr = 176400   => resonance shape f/fc invariant to fs.
  - per-cell phase step = 2*pi/(os*sr); cell k <-> freq = k/os Hz (0.25 Hz steps).
  - Phase-1 model FS=44100 was a valid normalization (res(f) depends only on f/fc).
 - case8 generator call (trace 33361..33490, switchD_1805318ab):
  - +0x814 numBands = 1, +0x103 mode flag = 1
  - param_5 (float, xmm4) = 10^(sens_dB/20)  (pow10 10.0^([band+0x101]/20.0))
  - gain = sqrtf(param_5) = 10^(sens_dB/40); model fit gain=4.132 => sens_dB@0x101 ~= 24.65
  - freq = [band+0x804], Q = [band+0x80c], fs_total = os*sr  (w0 = max(fc,2)*2pi/fs_total)
-## PHASE-2 GATE: PASSED (2026-08-18)
+## PHASE-3 Caller + Grid
 - `dsp/twin.hpp` + `dsp/twin.cpp` written: build_twin_coeff (FUN_180533ec0 double pipeline
  -> cvtpd2ps), cplx_div_exact (0x181a77520: mulps |A|^2, rcpps + Newton, NaN guard),
  cplx_mul_exact (0x18000ad60 scalar body @0x18000ae00: vfmaddsub213ps form),
  twin_apply (seed A0/B0 + 2 vfmadd213ss Horner stages + 2*div).
 - twin_check gate: fc-scan grid (7 fc, Q=0.9999978, gain=4.132, tone 1000 Hz)
  max |rel err| = 1.27e-5 vs double reference res_at (float rcpps parity).
  DUAL sweep (Q=0.1..10 x tones 500/2000, fc=500) max = 1.23e-5. Both PASS.
 - Constants locked: A1 = B1 = cos(w0)*-2.0, z1 = conj(exp(+i*theta)),
  z2 = cplx_mul(z1,z1), out = 2*B/A.
-## Confirmed scalar math (from generator + cplx-div kernels)
+- Caller `FUN_180536300` (6218): scratch 52·N, `0x18052da00` ramp `2π/(os·sr)`, `0x18052dbc0` interleave `e^{+iθ}`. Twin всегда `180535880`, sibling `180536f90` dead.
- Generator FUN_180533ec0 (per band, case8):
+- Grid: `os=4` (`+0x240080`), `sr=44100` (`+0x24`), `fs_total=176400`, `phase=2π/176400`, cell `k ↔ k/4 Hz` (0.25 Hz).
-  `fVar1 = sqrtf(param_5)`  (param_5 = pow(10, sens/20) via cd6 IAT stub)
+- case8: `+0x814=1`, `+0x103=1`, `param_5=10^(sens/20)`, `gain=√param_5=10^(sens/40)` → fit 4.132 ⇒ `sens≈24.65 dB`.
  `w0 = max(2.0, freq)*2*pi / fs_total`   (fs_total = 2*FS frame?)
  `p  = sin(w0)*0.5 / Q`
  `dVar5 = p*fVar1` (A tap), `param_4 = p/fVar1` (B tap)
  A = [1+dVar5, dVar2, 1-dVar5],  B = [1+param_4, dVar2, 1-param_4]
  where dVar2 = cos(w0) * DAT_1824c46b0  (=-2.0? VERIFY)
 - Twin FUN_180535880: for each FFT bin z (rotor):
  `res(z) = 2 * B(z) / A(z)`,  B(z)=B0+B1*z+B2*z^2, A(z)=A0+A1*z+A2*z^2
  implemented as complex FMA accumulation over 3 taps:
    r15 (A-accum) and r12 (B-accum) via 0x180001fd0 vfmadd213ss (coeff [rsp+k*4+0x28]=A_k, [+0x38]=B_k)
    rbp buffer = z powers via rotor ops (0x1800019a0 fill / 0x1800021e0 / 0x1800018b0 / 0x180001af0 / 0x180001dc0)
    final 0x180001e20 -> 0x181a775a0 cplx-div: out = 2*B/A  (xmm7=0x40000000=2.0, xmm6=NaN guard)
 - cplx-div kernel 0x181a77520: |A|^2 via shufps 0x88/0xdd, rcpps + 1 Newton step (xmm7=2.0), NaN guard via cmpeqps.
  Matches rotor_kernel.hpp cplx-div exactly.
-## Dispatch (all stubs share global 0x1826159a0 = 4 = AVX2)
+## PHASE-2 Gate
  0x180001d00 -> 0x180009860 -> 0x180040d40  (cplx fill const, NT stores)
  0x1800019a0 -> 0x180004200                 (float fill -> 0x181a63fe0)
  0x1800021e0 -> 0x1800130e0                 (large AVX, 0x450 stack, rotor/exp?)
  0x1800018b0 -> 0x180003040                 (-> 0x181a557c0)
  0x180001af0 -> 0x180005a20                 (loop -> 0x18001a5a0)
  0x180001dc0 -> 0x18000ad60                 (cplx op r14*rbp)
  0x180001e20 -> 0x18000b820                 (-> 0x181a775a0 cplx-div 2B/A)
  0x180001fd0 -> 0x18000f620                 (vfmadd213ss: out = src*coeff + dst)
-## Coefficient layout in twin (param_3 = {double A[3], double B[3]})
+- `dsp/twin.cpp`: `build_twin_coeff` (FUN_180533ec0), `cplx_div_exact` (rcpps+Newton), `cplx_mul_exact` (vfmaddsub), `twin_apply` `2·B/A`.
-  stack: +0x28=A0 +0x2c=A1 +0x30=A2 | +0x38=B0 +0x3c=B1 +0x40=B2
+- Gate: fc-scan 7fc + dual Q0.1..10 PASS.
  loop k=0..2 uses A_k via [rsp+k*4+0x28], B_k via [rsp+k*4+0x38]
  r15 = out A-accum, r12 = out B-accum, rbp = z^1..z^2 buffer, r14 = rotor(z) input
-## PHASE-1 LOCK (empirical, 36 points, LUT chain B.11)
+## Scalar math
-  Effective sqrtf(param_5) needed ~= 4.13 (NOT 1.995).
+
-  => param_5 ~= 17.07 = 10^(1.2325)  => internal sens_stored ~= 24.65 dB (not XML 12.0)
+- Generator: `fVar1=√param_5`, `w0=max(2,freq)·2π/fs_total`, `p=sin(w0)·0.5/Q`, `A=[1+p·fVar1, −2cos w0, 1−p·fVar1]`, `B=[1+p/fVar1, −2cos w0, 1−p/fVar1]`.
-  OR equivalent. RULED OUT: sqrt(10^(12/20))=1.995 (rmse 1.38/5.68);
+- Twin: `res(z)=2·B(z)/A(z)`, Horner `0x180001fd0`, `cplx-div 0x181a77520`, `z2=z1·z1`.
-  10^(12/20)=3.981 direct (t1kq 0.056 but t1k 0.934); |2B/A|^2 (1.5/6.6).
+
-  Sens setters (0x18053795d, 0x18056f002) both store RAW value; +0x808 scaling is HOST-side
+## Dispatch
-  (plugin controller, outside dump) => must be resolved via bit-exact render or host trace.
+
-  NOTE: single-gain compromise 4.132; t1kq wants 4.055, t1k wants 4.137 (tilt residual).
+Стабы → `0x1826159a0=4` (AVX2). Layout `param_3={A[3],B[3]}` `+0x28/+0x38`, coeff `[rsp+k*4]`.
 ## PHASE-1 lock
 `√param_5≈4.13` (не 1.995), `sens≈24.65 dB` (host ×2 над XML 12.0). Setter `+0x808` host-side.
 Детали декомпа — `handoff/nls_dasm/twin*.dis`, `dsp/twin.cpp:27`.
@@ -1,254 +1,8 @@
-# SESSION HANDOFF — soothe2 detector reverse → bit-exact render
+# SESSION_HANDOFF — перенесён в архив
-Prepared: 2026-08-18 (checkpoint end-of-session: commits c8f97e4 + 60bf3a2 pushed). Start: READ THIS FIRST.
+> **Исторический чекпоинт 2026-08-18 (коммиты c8f97e4+60bf3a2).**
 > Актуально: декомп DSP-ядра ~95% закрыт, `FUN_180529fe0` mono-path, `FUN_180563440/563ce0/563a60` расшифрованы, дизассемблы в `handoff/nls_dasm/` (134 файла).
 > Полный текст сохранён: [`handoff/archive/SESSION_HANDOFF_2026-08-18.md`](archive/SESSION_HANDOFF_2026-08-18.md) (256 строк, инвентарь декомпа §0, ключевые адреса §2, Phase-5 план §6).
 > Навигация → [`handoff/NOTES_LEVEL_INDEX.md`](NOTES_LEVEL_INDEX.md) + [`handoff/BLOCKMAP_529fe0.md`](BLOCKMAP_529fe0.md) + [`README.md:13`](../README.md).
-## 0. DECOMPILATION INVENTORY (2026-08-19 — what's decoded, where, and what's missing)
+Кратко: `decomp_funs.txt` 312K строк, RTTI `rtti_dsp.json`, bridge `C=g·LUT+w·warp^a` (rmse 0.236, позже заменён VLAW). Для продолжения читать `AGENTS.md` → `NOTES_LEVEL_INDEX.md`.
 Goal: bit-exact parity is gated by EXACT tables/window/constants. The chunk-level model hits
 err ≤0.62 dB (dual) / ≤0.19 dB (al_*) — to go sample-exact we need precise values from the binary.
 ### Static decomp assets IN REPO (use these, don't re-decompile):
 - `ghidra-proj/soothe2.rep` — full Ghidra project (vst3 at ImageBase 0x180000000).
 - `decomp_funs.txt` (312K lines, ~1640 functions), `fun_map.txt` (2285 addr→FUN), `consts.txt` (11602),
  `decomp_dsp.txt` / `decomp_vtables.txt` / `decomp_candidates.txt` / `focus_decomp.txt`. Generators:
  `Dump*.java`, `ImportRtti*.java`, `ListFuns.java`, `SearchRefs.java` (+ headless logs).
 - RTTI: `rtti_dsp.json` / `rtti_full.json` — class hierarchy
  `SpectralProcessor<float,7,1>`, `Soothe2ModuleBase<float,1>`, `FilterGraph<float,6,0x400>`,
  `DigitalFilter<float,0xBA,1>`, `IIRFilterExtended<float,1>`, `AudioProcessingModule<float,1>`.
 - `handoff/nls_dasm/` — 120 hand-picked `.dis` (twin, iface_18052da00/dbc0, fft, generator, ctor).
 - `dsp/` — working C++ transcription (twin, detect, freqpath, spectral, fft_stage) + harness;
  `build/twin_check` passes float-parity gate (§2).
 ### KEY DSP ADDRESSES — decoded / not decoded:
 DECODED (formula-level, notes at NOTES_TWIN.md / NOTES_LEVEL.md):
 - twin kernel `FUN_180535880`; generator `FUN_180533ec0`; caller `FUN_180536300`; dead sibling `180536f90`.
 - level-weight formula `FUN_180530d30` (0x540880/884, warp 0x5406a8=0.87·x/(1+x/7.942), w=0.1^(...)).
 - mask-apply entry `FUN_180529fe0` (accumulator 0x5407c8 += w·res; mask *= warp; FIR *= 0x540658).
 - FFT-conv loop 0x52b550-0x52b8b5 (plan 0x540530, windows 0x540548/550/598, freq-axis 0x540698=offline const).
 - level-path map: `0x563440` (LUT curve +0x188, 6 band-slots, combine→+0x2198), `0x56e3e0` (twin-mask factory),
  `0x563ce0` (IIR level-tracker INIT only).
 NOT DECODED / MISSING FROM decomp_funs.txt (critical):
 - `FUN_180529fe0` body — mask/fir apply; only reachable via runtime + vtable slot `180529fe0` in fun_map.txt.
 - `FUN_180563440` (LUT curve + gamma + combine), `FUN_180563ce0` (IIR level-tracker UPDATE loop).
  Full disasms exist ONLY in `/tmp/opencode/f_563440.dis`, `f_563ce0.dis`, `f529fe0.dis` — **/tmp is
  ephemeral; COPY INTO handoff/nls_dasm/ on next session start.**
 - Window `0x540658` (single indirect write — statically invisible); live-captured copies saved as
  `rwin_A0/A1/B0/C0.npy` + `r_freqaxis.npy` (see NOTES_LEVEL.md §2026-08-18h3). Live data at SR=48000
  (offline renders 44100 → renormalize warp x by actual Nyquist).
 - sens source: XML 12.0 → runtime sens_dB≈24.65 (host ×2) not found in dump; `IAT\*0x181bab370` outside dump.
 ### Bridge model STATUS (2026-08-19):
 - `framed_render.py` full STFT frame-render (N=2048, hop=512, sqrt-Hann, twin env tatt=11ms/trel=80ms):
  `C(f)=G·LUT(xv)+W·warp(f)^A`, G=1.0850, W=0.2819, A=1.1377 (joint dual+al_* refit).
 - Validation: dual q0.1/1/10 @500+2000 err ≤0.62 dB, envRmse@steady ≤0.78 dB; al_* lv3..24 err ≤0.19 dB.
 - LUT = slanted al_*-leg (0.366@xv=-0.50 → 0.636@xv=+0.55), replaces flat B.12 (~0.5). Bugfix: clip range
  must be [LY.min(), LY.max()] not [LY[0], LY[-1]].
 - Remaining structural residual: −0.6 dB systematic on dual 500Hz (q0.1/q1). Live evidence (avg per-bin
  gain H=|Y|/|X|): reference mask is FLAT ~-10.2dB across 100-540Hz regardless of Q — model produces
  res-shaped notch. Hypo tested: freq-smoothing of C(f) fails (kills 2000Hz). NEXT: scalar per-frame xv
  (broadband level, not per-bin am/res) — see /tmp/smoothtest.py (+ edit scalar=True).
 ## 1. Objective (unchanged since session 1)
 Transcribe the decoded soothe2 detector ("twins" 0x180535880/0x180536f90, 2nd-order resonator) into
 C++ and reach **bit-exact render parity** with the Reaper reference wavs in `/home/m/soothe-bt/*.wav`
 (checking dB/magnitude parity first, then byte-diff).
 Path A (bit-exact) chosen. Phase 0,1,2 done. Phase 3,4 advanced. Phase 5 (render diff) is next.
 ## 2. DECOMPILATION FINDINGS (2026-08-19 — FUN_180563440, FUN_180563ce0 decoded)
 ### FUN_180563440 — LUT curve evaluation + band combine (222 lines disasm)
 Structure: **3 phases per frame**:
 1. **1024-bin LUT loop** (0x400 iterations): for each bin k:
   - x = clamp(k * 0.0009775, 0, 1.0) = k/1024
   - Band config at +0x188: {A(+0x00), B(+0x04), threshold(+0x0c), flag(+0x10), callback(+0x50)}
   - **Path 1** (callback exists): vtable call → dynamic LUT
   - **Path 2** (flag=1, threshold≠1.0): **power-law** → centered = 2*x - 1, then `sign(x) * 10^(log10(|x|) / threshold)` — this is a **compression curve** controlled by sharpness/threshold
   - **Path 3** (default): **linear interpolation** → `(B - A) * x + A`
   - Output: double-precision at +0x198, stride 8
 2. **Twin-mask factory** (FUN_18056e3e0): 6 bands × 1024 bins, stride 0x2000
 3. **Combine loop**: stereo (max 2 channels), 6 bands, `1 - sum(band_masks)`
 ### FUN_180563ce0 — IIR level-tracker INIT (163 lines disasm)
 - **341 bins** (0x155 iterations), **order-3 IIR** (3 coefficients per bin)
 - Coefficient: **0.1** (`0x3dcccccd` = IEEE 754 float 0.1)
 - Initial state: [1.0, 0, 0, 0] and [-1.0, 0, 0, 0] (identity + zero)
 - Buffer layout: 3 × (16 bytes coeff) per bin, stored at rcx+0x28/+0x40/+0x58
 - **Not the update loop** — init only; UPDATE is elsewhere
 ### FUN_180529fe0 — Coefficient setup (2051 instructions, in decomp_funs.txt)
 - **vtable method** on Soothe2Module<M,1>
 - **Lock** at +0x2404dc (atomic test-and-set)
 - **PRNG state** at +0x2404e0, LCG with offset 0x3cdca
 - **6-iteration coefficient generation** from 0x5408b0 buffer (LCG-indexed)
 - **Depth scaling**: `powf(normalized, depth)` at +0x2c
 - **Mask assembly**: normalize by 0x1a0 (NFFT), × 0x540870 (level weight), × 0x54088c (sharpness), invert
 - **Copy output** via SIMD memcpy (thunk 0x181ba94b0)
 ### FUN_18052e9b0 — SpectralProcessor main (3167 instructions, in decomp_funs.txt)
 - **Same PRNG + coefficient setup** as FUN_180529fe0
 - **Band chain**: FUN_18052f500 (interleave) → FUN_18052ee70 (per-bin gain) → FUN_18052d650 (setup) → FUN_18052d920 (window)
 - **Buffer alloc**: FUN_18052e190 for 0x540668, 0x540698, 0x5406a8, 0x5406b8-e8 (6 bands)
 - **Depth scaling**: `powf(normalized, depth)` with depth at +0x2c
 - **Final mask**: `1 - C` (inversion)
 - **Window application**: `FIR *= 0x540658` (the live-captured window table)
 ### Constants verified from binary:
 | Constant | Address | Value | Meaning |
 |---|---|---|---|
 | SCALE | 0x24c3c54 | 0.000977517 | 1/1024 (bin→x) |
 | ONE | 0x24c3ea4 | 1.0 | clamping max |
 | TWO | 0x24c41e0 | 2.0 | centering (2*x-1) |
 | NEG1 | 0x24c4680 | -1.0 | sign flip |
 | HALF | 0x24c3d8c | 0.5 | threshold |
 | DEPTH_SCALE | 0x24c4334 | 4.0 | depth range |
 | DB_CONV | 0x24c43e0 | 8.6859 | 20/ln(10) |
 | FLOOR | 0x24c4704 | -6.9078 | ln(0.001) |
 | IIR_COEFF | embedded | 0.1 | attack/release per bin |
 | IIR_ORDER | embedded | 3.0 | IIR filter order |
 | LCG_OFFSET | embedded | 0x3cdca | PRNG state advance |
 ### Live-captured tables (re-verified):
 - `rwin_A0.npy` (0x1930100): 0.5→0.8, **frequency window** (NOT warp formula)
 - `rwin_B0.npy` (0x1938180): 0→3.899, **power-law depth curve** (exponent ~0.66)
 - `rwin_C0.npy` (0x19401c0): 0.596→0.126, **level-dependent weight**
 - `r_freqaxis.npy`: 0→23988.3 Hz, 11.71 Hz spacing (48000/4096)
 ### Key insight: The warp formula is NOT a table — it's computed at runtime
 The empirical `0.87*7.942*x/(7.942+x)` is an approximation of a runtime computation
 in FUN_180563440 that evaluates the LUT curve parametrically. The actual LUT has TWO modes:
 - **Linear** (default): simple interpolation between A and B
 - **Power-law** (flag=1): `sign(x) * 10^(log10(|x|) / C)` — compression curve
 ## 3. PROOF OF STATE — run this first (5 min)
 Everything below must reproduce. If `twin_check` fails, the transcription moved stale.
 ```bash
 cd /home/m/re-tools/dsp && cmake --build build && ./build/twin_check
 # expect:
 #   max rel err = 1.268e-05, PASS = yes (float-parity)
 #   max sweep rel err = 1.233e-05, SWEEP PASS = yes, exit 0
 ```
 `build/soothe2_dsp.so` and `build/twin_check` are current. CMakeLists already has `twin.cpp` + `twin_check` executable.
 ## 3. TRANSCRIPTION STATE (summary — details in NOTES_TWIN.md)
 - `dsp/twin.hpp` + `dsp/twin.cpp` = twin kernel FUN_180535880:
  - build_twin_coeff (FUN_180533ec0 double pipeline → cvtpd2ps floats)
  - generator formula (confirmed): `fVar1=sqrtf(param_5)`; `w0=max(2.0,freq)·2π/fs_total`;
    `p=(sin(w0)·0.5)/Q`; `A=[1+dVar5, dVar2, 1−dVar5]`, `B=[1+param_4, dVar2, 1−param_4]`,
    `dVar2=cos(w0)·−2.0` (DAT_1824c46b0=−2.0).
  - cplx_div_exact (rcpps+Newton: `ref=r0·(2−den·r0)`, NaN-guard cmpeqps+movmskps, numerator B·conj(A))
  - cplx_mul_exact (vfmaddsub213ps form `re=fma(br,ar,−bi·ai)`, `im=fma(br,ai,bi·ar)`)
  - twin_apply: conj(z1) mirroring 0x1800018b0, z2=z1·z1, out=2·B/A
 - **OPEN phase question (byte-exactness)**: conjugate shaping — (-α+π/2) vs (-α) axis. Magnitudes match
  either way; final proof needed via phase of a single-bin run before claiming byte-exactness.
 ## 4. GRID + CALLER (Phase 3 — details in NOTES_TWIN.md §Phase 3)
 - Caller FUN_180536300 (decomp 6218, `caller_536300.dis`), scratch ≥52·N complex bytes:
  +0..8N A-accum, +8N..16N B-accum, +16N..24N z-powers, +24N..32N rotor-in (filled 1.0),
  +32N..36N phase ramp lVar1, +36N..44N rotor-out z1, +44N..52N lVar3(1.0)/lVar4(0.0)
 - `0x18052da00(lVar1, scale=2π/(os·sr), N)` = ramp fill; `0x18052dbc0(param_2, lVar4, N)` cplx-interleave
  → out slots get (cos,sin) = e^{+iθ}.
 - **Grid law**: `os=*(int*)(cfg+0x240080)=4` (mov DWORD PTR [rbx+0x240080],0x4 @0x18052ce3e),
  `sr=*(float*)(cfg+0x24)=44100.0` → generator `fs_total=os·sr=176400`; per-cell phase `2π/176400`;
  cell k ↔ freq k/os Hz (0.25 Hz steps). Phase-1 FS=44100 was a valid normalization (res depends on f/fc).
 - twin always = FUN_180535880 (type-info compare equal); sibling FUN_180536f90 dead.
 - case8: numBands `+0x814`=1, `+0x103`=1; `param_5=10^(sens_dB/20)`, `gain=sqrtf(param_5)=10^(sens_dB/40)`
  → fit gain=4.132 means stored `sens_dB@+0x101 ≈ 24.65`; freq=[band+0x804]=1000.0, Q=[band+0x80c].
 ## 5. LEVEL-WEIGHT FORMULA (Phase 4 — DECODED, NEGATIVE RESULT — NOTES_LEVEL.md §UPDATE 2026-08-18)
 Exact formula from decomp 22330 (FUN_180530d30):
 - `base=(2000/(sr·0.5))·iVar5/(bin+1)`, iVar5=NFFT/2+1
 - `w8=powf(base,0.25)`
 - `v=[0x540880]·0.25·w8·fVar12` (fVar12=4.0 if offline flag [0x5408b8], else 1.0)
 - `q=1/(1+v/([0x540880]·4))` (2nd pair 0x540884: q2=1/(1+v2/[0x540884]))
 - `dVar1=(sr/[0x1a0])·[0x1ac]·0.001`
 - `w=0.1^(1/(max(q·v,floor)·dVar1))`; bufs: 0x5406b8=w, 0x5406c8=1−w (and 0x5406d8/6e8 w/ 0x540884)
 - consts: c3d3c=0.25, c4334=4.0, c3d8c=0.5, c45b4=2000.0, c3f70=0.1 f32, c3e30=0.001 f64, c3f70=0.1
 - **Numerically**: w(500)=w(1000)=w(2000)≈0 (1e-20..1e-0.07), complements≈1 → but the weights are STILL
  APPLIED (as 1−w ≈ 1) by FUN_180529fe0 (`0x5407c8[band] += 0x5406c8·res_upper + 0x5406e8·res_lower`,
  thunk 003c40, kernel 0x1800752e0 = vfmadd213pd) — the form is weight·resonance, not dead code.
 - **TILT analysis (2026-08-18, checkpoint NEGATIVE result)**: warp `0x5406a8 = 0.87·x/(1+x/K)`,
  `K=exp(2.0723267)≈7.942`, `x=f/2000`, applied by FUN_180529fe0 as `mask *= 0x5406a8` (float, thunk 008700)
  BEFORE the FFT-conv. warp: 500→0.211, 1000→0.409, 2000→0.773 (lin ratio 2000/500=3.67).
  **BUT warp ≠ empirical tilt (ratio 1.27)**: `C(dB2000)=depth·warp(2000)·LUT_max = 0.864·0.772·0.667=0.445`,
  yet red2000=15.2 dB needs `C=0.826` ⇒ warp alone CANNOT produce deep-2000 (dual_b1q). Empirical tilt is the
  envelope of the BLEND term (0x5407c8 += weights·res + freq-axis 0x540698) + FFT-conv shaping, NOT warp —
  simple tilt→warp substitution in model_lut.py is INVALID. Nothing else decoded yet closes dual_b1q.
 - **FUN_180563ce0 decoded (2026-08-18)**: INIT of per-bin IIR level-trackers (0x156=342 bins, order=3.0f
  0x40400000, coeff A/B from 0x24c4780/4790, strided fill; third pass +0xffd0 fills 0x40e00000=7.0f).
  Not the update loop. Full disasm: `/tmp/opencode/f_563ce0.dis`.
 - **Level-path map (2026-08-18, NOTES_LEVEL.md)**: 0x563440 (LUT curve +0x188, gamma, 6 band-slots, combine
  →+0x2198, LUT pairs +0x98 at 0x3ff pts) + 0x56e3e0 twin-mask factory (N=1024, DOUBLE, LUT-ramp input) +
  0x563ce0 IIR init. 0x563440 NOT in decomp_funs.txt (search via /tmp/opencode/f_563440.dis + level_notes.md).
 - **DIAGNOSTIC BRIDGE DONE (2026-08-18) — model_fir.py**: реальная цепочка из FUN_180529fe0
  воспроизводит 36 точек до rmse=0.236 dB БЕЗ эмпирического tilt:
    `C(f) = g·LUT(log10(L0/res(f))) + w·warp(f)^a`, g=1.221, w=0.358, a=3.143.
  - dual500-константа = res_band(500;fc=500)=0.117 Q-НЕЗАВИСИМ, не tilt;
  - dual2000-глубина = АДДИТИВНЫЙ терм `w·warp^3.14` (на 1000 Гц вклад ≤0.021 — иначе t1k
    рушится; на 2000 = 0.149) — это аккумулятор `0x5407c8 += weights·res + mask` (шаг 5),
    НЕ мультипликация warp·LUT (та проваливается >10 dB);
  - warp^~3.14 ≈ π — подозрение на кратный каскад (0x540698 freq-axis, ∏0x540688, двойной FFT);
  - остаток 0.7 dB при Q=0.1 = форма LUT-колена 0.574, не закрывается per-bin → нужен FFT-уровень.
 - **FFT-CONV FULLY MAPPED (2026-08-18, NOTES_LEVEL.md §2026-08-18c, /tmp/opencode/f_52b570.dis)**:
  0x535a70 = dispatcher to base-CRT FFT (0x140a10/0x140a70); 0x540548/0x540550/0x540598 = FFT-plan
  tables (plan 0x540530, built by 0x52dc30), NOT windows. Per-band loop 0x52b550-0x52b8b5:
  mask→fwdFFT→invFFT→fill low/high (xmm13/xmm9 consts)→inv→cplx-op→fwd→`FIR *= 0x540658`→fill→inv→
  `FIR[0]=1,FIR[1]=0`→`*=0x540888`→copy out.
  **freq-axis 0x540698**: OFFLINE = constant scalar (NOT exp-formula — exp/rand = online dither).
  **Window 0x540658**: only ONE direct ref in dump (read at 0x52b771), written only indirectly —
  STATICALLY INVISIBLE → the only plausible source of warp^3.14/LUT-knee. Need runtime capture or
  full-pipeline empirical closure.
 ## 6. PHASE 5 PLAN (updated 2026-08-18 — bridge done, tilt replaced, FFT mined out)
 Decision (user): bridge FIRST. **DONE — bridge result above in §5.**
 Status: step 2 DONE, step 3 PARTIAL (static analysis exhausted).
 1. (15 min) Proof-of-state above; confirm twin + mdl files complete; repo at `60bf3a2`.
 2. **Replace empirical tilt** in model_lut.py with `g·LUT(xv) + w·warp(f)^a` — **DONE**, rmse 0.236 dB
   reproduced, committed `c8f97e4`. Optional refinement fit (2 LUT-knots) reaches 0.167 but is overfit.
 3. **Close the 0.7-dB gap**: FFT-conv loop mapped + freq-axis cleared (offline const) — **DONE/PARTIAL**,
   committed `60bf3a2`. Remaining unknown = 0x540658 window content (static-invisible). Options:
   (a) runtime capture of 0x540658, (b) skip and close empirically in full-pipeline step 5.
 4. Runtime values: per-bin IIR level trackers (0x563ce0), 0x540880/884, `[0x1a0]/[0x1ac]`, cell count N.
 5. Full pipeline + render diff vs `/home/m/soothe-bt/*.wav`: dB parity → byte parity.
 NEXT SESSION START (recommendation): skip deeper FFT mining; go straight to step 5 full pipeline
 (mask-chain already in model_lut.py B.12 at 0.236; freq-axis const; FFT-conv reduces to a per-bin
 smoothing + 0x540658 shaping that can be fit/absorbed). Runtime capture of 0x540658 only if step 5
 stalls on the Q=0.1 knee.
 ### STEP 5 STARTED (2026-08-18) — dB-parity harness DONE (commit f96bb42)
 - `/home/m/re-tools/render_parity.py`: measures steady-state per-tone reduction DIRECTLY on the
  reference wavs (Goertzel tone-amp over late steady window; handles 16/24-bit, mono/stereo).
  NOTE: Goertzel must use the SEQUENTIAL update form (s2=s1;s1=s0;s0=v+cw*s1-s2) — the tuple-unpack
  form is numerically unstable (overflows ~1500 steps).
 - Validation: measured dual Q-scan == stored dataset (dual.wav is 4s mono 16-bit; renders 6s stereo 24-bit).
 - **dB-PARITY RESULT (canonical B.12, 36 pts, real wavs): TOTAL rmse=0.268 dB**
  (dual500 0.130, dual2000 0.387, t1kq 0.320, t1k 0.076).
  residual structure: dual2000 -0.80@Q0.1..+0.36@Q10 (LUT-knee), t1kq +0.36..+0.26 (level region),
  t1k ±0.1 (good). Refit (g,w,a)=(1.224,0.379,3.405)→0.264; +2 LUT-knots→0.203 (plateau).
 - CONCLUSION: B.12 form plateau ~0.20-0.27 dB vs real renders; residual is structural (0x540658
  FFT shaping + level-curve region), matches plan expectation. Re-measured-on-wav numbers differ
  slightly from the stored model_lut.py dataset (t1kq ~+0.3 dB) — different steady-window method.
 - Open: full PIPELINE frame-render (STFT + per-frame mask + time constants) is the remaining big
  lift for byte-parity; dB-parity milestone achieved at 0.268 dB.
 Candidates that may consume Phase 5 time: source of sens≈24.65 (host ×2 over XML 12.0 — not found in dump),
 IAT\*0x181bab370 outside dump (non-blocking), final z¹ phase proof, the 0x563440 LUT curve (+0x188) values.
 ## 7. FILES LAYOUT
 - `/home/m/re-tools/dsp/` — twin.{hpp,cpp}, twin_check.cpp, CMakeLists.txt, build/
 - `/home/m/re-tools/model_lut.py` — model **B.12** (Q=xmlq, gain=10^(sens/20), depth 0.8639736175537109,
  `C=g·LUT(xv)+w·warp^a`, g=1.221/w=0.358/a=3.143) — 0.236 dB bridge model; PCHIP LUT nodes frozen
 - `/home/m/re-tools/model_fir.py` — bridge canonical source (rmse 0.236, committed b1066f3)
 - `/home/m/re-tools/framed_render.py` — full frame-render pilot (2026-08-19 params; modes dual|al)
 - `/home/m/re-tools/rwin_A0/A1/B0/C0.npy`, `r_freqaxis.npy`, `rwin_warp.npy` — live-window tables (48k)
 - `/home/m/re-tools/rtsnap.py` — live-process page snapshotter; snapshots /tmp/rt{A,B,C,D}.{raw,idx}
 - `/home/m/re-tools/handoff/decode_rpp_full.py`, `handoff/rpp_allparams.py` — RPP b64-XML full decoder
  (trim to len%4==0, `<?xml`@92, incl. nested processorStateData)
 - `/home/m/re-tools/soothe_mem.bin` — memory dump, offset=VA−0x180000000
 - `/home/m/re-tools/decomp_funs.txt` — decomp: FUN_180536300 (6218), FUN_180530d30 (22330),
  FUN_180533ec0 (123204), FUN_180529fe0 (missing body! vtable-only)
 - `/home/m/re-tools/handoff/` — THIS DOC, NOTES_TWIN.md, NOTES_LEVEL.md, phase1/*.py, nls_dasm/*.dis
 - `/home/m/soothe-bt/*.rpp` (decoder `/home/m/re-tools/dsp/...`; also `decode_rpp4.py` referenced) + `*.wav` refs
@@ -0,0 +1,256 @@
 # SESSION HANDOFF — soothe2 detector reverse → bit-exact render
 Prepared: 2026-08-18 (checkpoint end-of-session: commits c8f97e4 + 60bf3a2 pushed). Start: READ THIS FIRST.
 > **2026-08-20 UPDATE**: см. актуальный канон — `AGENTS.md` и `handoff/NOTES_LEVEL.md`.
 > Нижеследующее «NOT DECODED»/«missing body» УСТАРЕЛО: `FUN_180529fe0` mono-path,
 > `FUN_180563440/563ce0/563a60` — расшифрованы, полные дизассемблы скопированы в
 > `handoff/nls_dasm/` (134 файла). Этот файл — исторический чекпоинт.
 ## 0. DECOMPILATION INVENTORY (2026-08-19 — what's decoded, where, and what's missing)
 Goal: bit-exact parity is gated by EXACT tables/window/constants. The chunk-level model hits
 err ≤0.62 dB (dual) / ≤0.19 dB (al_*) — to go sample-exact we need precise values from the binary.
 ### Static decomp assets IN REPO (use these, don't re-decompile):
 - `ghidra-proj/soothe2.rep` — full Ghidra project (vst3 at ImageBase 0x180000000).
 - `decomp_funs.txt` (312K lines, ~1640 functions), `fun_map.txt` (2285 addr→FUN), `consts.txt` (11602),
  `decomp_dsp.txt` / `decomp_vtables.txt` / `decomp_candidates.txt` / `focus_decomp.txt`. Generators:
  `Dump*.java`, `ImportRtti*.java`, `ListFuns.java`, `SearchRefs.java` (+ headless logs).
 - RTTI: `rtti_dsp.json` / `rtti_full.json` — class hierarchy
  `SpectralProcessor<float,7,1>`, `Soothe2ModuleBase<float,1>`, `FilterGraph<float,6,0x400>`,
  `DigitalFilter<float,0xBA,1>`, `IIRFilterExtended<float,1>`, `AudioProcessingModule<float,1>`.
 - `handoff/nls_dasm/` — 120 hand-picked `.dis` (twin, iface_18052da00/dbc0, fft, generator, ctor).
 - `dsp/` — working C++ transcription (twin, detect, freqpath, spectral, fft_stage) + harness;
  `build/twin_check` passes float-parity gate (§2).
 ### KEY DSP ADDRESSES — decoded / not decoded:
 DECODED (formula-level, notes at NOTES_TWIN.md / NOTES_LEVEL.md):
 - twin kernel `FUN_180535880`; generator `FUN_180533ec0`; caller `FUN_180536300`; dead sibling `180536f90`.
 - level-weight formula `FUN_180530d30` (0x540880/884, warp 0x5406a8=0.87·x/(1+x/7.942), w=0.1^(...)).
 - mask-apply entry `FUN_180529fe0` (accumulator 0x5407c8 += w·res; mask *= warp; FIR *= 0x540658).
 - FFT-conv loop 0x52b550-0x52b8b5 (plan 0x540530, windows 0x540548/550/598, freq-axis 0x540698=offline const).
 - level-path map: `0x563440` (LUT curve +0x188, 6 band-slots, combine→+0x2198), `0x56e3e0` (twin-mask factory),
  `0x563ce0` (IIR level-tracker INIT only).
 NOT DECODED / MISSING FROM decomp_funs.txt (critical):   ⚠️ → РЕШЕНО (2026-08-19/20), см. выше
 - `FUN_180529fe0` body — РЕШЕНО (decomp в `/tmp/consumers_out.txt`, mask-цепь в `dsp/framed_model.cpp`).
 - `FUN_180563440` / `FUN_180563a60` (LUT curve + gamma + combine), `FUN_180563ce0` — РЕШЕНО,
  дизассемблы скопированы в `handoff/nls_dasm/` (`f_563440.dis`, `f_563a60.dis`, `f_563ce0.dis`, `f529fe0.dis`).
 - Window `0x540658` — РЕШЕНО (live-захват, см. `handoff/NOTES_CAPTURE.md`; `rtwin_freq_44100.npy`).
 - sens source: XML 12.0 → runtime sens_dB≈24.65 (host ×2) not found in dump; `IAT\*0x181bab370` outside dump.
 ### Bridge model STATUS (2026-08-19):
 - `framed_render.py` full STFT frame-render (N=2048, hop=512, sqrt-Hann, twin env tatt=11ms/trel=80ms):
  `C(f)=G·LUT(xv)+W·warp(f)^A`, G=1.0850, W=0.2819, A=1.1377 (joint dual+al_* refit).
 - Validation: dual q0.1/1/10 @500+2000 err ≤0.62 dB, envRmse@steady ≤0.78 dB; al_* lv3..24 err ≤0.19 dB.
 - LUT = slanted al_*-leg (0.366@xv=-0.50 → 0.636@xv=+0.55), replaces flat B.12 (~0.5). Bugfix: clip range
  must be [LY.min(), LY.max()] not [LY[0], LY[-1]].
 - Remaining structural residual: −0.6 dB systematic on dual 500Hz (q0.1/q1). Live evidence (avg per-bin
  gain H=|Y|/|X|): reference mask is FLAT ~-10.2dB across 100-540Hz regardless of Q — model produces
  res-shaped notch. Hypo tested: freq-smoothing of C(f) fails (kills 2000Hz). NEXT: scalar per-frame xv
  (broadband level, not per-bin am/res) — see /tmp/smoothtest.py (+ edit scalar=True).
 ## 1. Objective (unchanged since session 1)
 Transcribe the decoded soothe2 detector ("twins" 0x180535880/0x180536f90, 2nd-order resonator) into
 C++ and reach **bit-exact render parity** with the Reaper reference wavs in `/home/m/soothe-bt/*.wav`
 (checking dB/magnitude parity first, then byte-diff).
 Path A (bit-exact) chosen. Phase 0,1,2 done. Phase 3,4 advanced. Phase 5 (render diff) is next.
 ## 2. DECOMPILATION FINDINGS (2026-08-19 — FUN_180563440, FUN_180563ce0 decoded)
 ### FUN_180563440 — LUT curve evaluation + band combine (222 lines disasm)
 Structure: **3 phases per frame**:
 1. **1024-bin LUT loop** (0x400 iterations): for each bin k:
   - x = clamp(k * 0.0009775, 0, 1.0) = k/1024
   - Band config at +0x188: {A(+0x00), B(+0x04), threshold(+0x0c), flag(+0x10), callback(+0x50)}
   - **Path 1** (callback exists): vtable call → dynamic LUT
   - **Path 2** (flag=1, threshold≠1.0): **power-law** → centered = 2*x - 1, then `sign(x) * 10^(log10(|x|) / threshold)` — this is a **compression curve** controlled by sharpness/threshold
   - **Path 3** (default): **linear interpolation** → `(B - A) * x + A`
   - Output: double-precision at +0x198, stride 8
 2. **Twin-mask factory** (FUN_18056e3e0): 6 bands × 1024 bins, stride 0x2000
 3. **Combine loop**: stereo (max 2 channels), 6 bands, `1 - sum(band_masks)`
 ### FUN_180563ce0 — IIR level-tracker INIT (163 lines disasm)
 - **341 bins** (0x155 iterations), **order-3 IIR** (3 coefficients per bin)
 - Coefficient: **0.1** (`0x3dcccccd` = IEEE 754 float 0.1)
 - Initial state: [1.0, 0, 0, 0] and [-1.0, 0, 0, 0] (identity + zero)
 - Buffer layout: 3 × (16 bytes coeff) per bin, stored at rcx+0x28/+0x40/+0x58
 - **Not the update loop** — init only; UPDATE is elsewhere
 ### FUN_180529fe0 — Coefficient setup (2051 instructions, in decomp_funs.txt)
 - **vtable method** on Soothe2Module<M,1>
 - **Lock** at +0x2404dc (atomic test-and-set)
 - **PRNG state** at +0x2404e0, LCG with offset 0x3cdca
 - **6-iteration coefficient generation** from 0x5408b0 buffer (LCG-indexed)
 - **Depth scaling**: `powf(normalized, depth)` at +0x2c
 - **Mask assembly**: normalize by 0x1a0 (NFFT), × 0x540870 (level weight), × 0x54088c (sharpness), invert
 - **Copy output** via SIMD memcpy (thunk 0x181ba94b0)
 ### FUN_18052e9b0 — SpectralProcessor main (3167 instructions, in decomp_funs.txt)
 - **Same PRNG + coefficient setup** as FUN_180529fe0
 - **Band chain**: FUN_18052f500 (interleave) → FUN_18052ee70 (per-bin gain) → FUN_18052d650 (setup) → FUN_18052d920 (window)
 - **Buffer alloc**: FUN_18052e190 for 0x540668, 0x540698, 0x5406a8, 0x5406b8-e8 (6 bands)
 - **Depth scaling**: `powf(normalized, depth)` with depth at +0x2c
 - **Final mask**: `1 - C` (inversion)
 - **Window application**: `FIR *= 0x540658` (the live-captured window table)
 ### Constants verified from binary:
 | Constant | Address | Value | Meaning |
 |---|---|---|---|
 | SCALE | 0x24c3c54 | 0.000977517 | 1/1024 (bin→x) |
 | ONE | 0x24c3ea4 | 1.0 | clamping max |
 | TWO | 0x24c41e0 | 2.0 | centering (2*x-1) |
 | NEG1 | 0x24c4680 | -1.0 | sign flip |
 | HALF | 0x24c3d8c | 0.5 | threshold |
 | DEPTH_SCALE | 0x24c4334 | 4.0 | depth range |
 | DB_CONV | 0x24c43e0 | 8.6859 | 20/ln(10) |
 | FLOOR | 0x24c4704 | -6.9078 | ln(0.001) |
 | IIR_COEFF | embedded | 0.1 | attack/release per bin |
 | IIR_ORDER | embedded | 3.0 | IIR filter order |
 | LCG_OFFSET | embedded | 0x3cdca | PRNG state advance |
 ### Live-captured tables (re-verified):
 - `rwin_A0.npy` (0x1930100): 0.5→0.8, **frequency window** (NOT warp formula)
 - `rwin_B0.npy` (0x1938180): 0→3.899, **power-law depth curve** (exponent ~0.66)
 - `rwin_C0.npy` (0x19401c0): 0.596→0.126, **level-dependent weight**
 - `r_freqaxis.npy`: 0→23988.3 Hz, 11.71 Hz spacing (48000/4096)
 ### Key insight: The warp formula is NOT a table — it's computed at runtime
 The empirical `0.87*7.942*x/(7.942+x)` is an approximation of a runtime computation
 in FUN_180563440 that evaluates the LUT curve parametrically. The actual LUT has TWO modes:
 - **Linear** (default): simple interpolation between A and B
 - **Power-law** (flag=1): `sign(x) * 10^(log10(|x|) / C)` — compression curve
 ## 3. PROOF OF STATE — run this first (5 min)
 Everything below must reproduce. If `twin_check` fails, the transcription moved stale.
 ```bash
 cd /home/m/re-tools/dsp && cmake --build build && ./build/twin_check
 # expect:
 #   max rel err = 1.268e-05, PASS = yes (float-parity)
 #   max sweep rel err = 1.233e-05, SWEEP PASS = yes, exit 0
 ```
 `build/soothe2_dsp.so` and `build/twin_check` are current. CMakeLists already has `twin.cpp` + `twin_check` executable.
 ## 3. TRANSCRIPTION STATE (summary — details in NOTES_TWIN.md)
 - `dsp/twin.hpp` + `dsp/twin.cpp` = twin kernel FUN_180535880:
  - build_twin_coeff (FUN_180533ec0 double pipeline → cvtpd2ps floats)
  - generator formula (confirmed): `fVar1=sqrtf(param_5)`; `w0=max(2.0,freq)·2π/fs_total`;
    `p=(sin(w0)·0.5)/Q`; `A=[1+dVar5, dVar2, 1−dVar5]`, `B=[1+param_4, dVar2, 1−param_4]`,
    `dVar2=cos(w0)·−2.0` (DAT_1824c46b0=−2.0).
  - cplx_div_exact (rcpps+Newton: `ref=r0·(2−den·r0)`, NaN-guard cmpeqps+movmskps, numerator B·conj(A))
  - cplx_mul_exact (vfmaddsub213ps form `re=fma(br,ar,−bi·ai)`, `im=fma(br,ai,bi·ar)`)
  - twin_apply: conj(z1) mirroring 0x1800018b0, z2=z1·z1, out=2·B/A
 - **OPEN phase question (byte-exactness)**: conjugate shaping — (-α+π/2) vs (-α) axis. Magnitudes match
  either way; final proof needed via phase of a single-bin run before claiming byte-exactness.
 ## 4. GRID + CALLER (Phase 3 — details in NOTES_TWIN.md §Phase 3)
 - Caller FUN_180536300 (decomp 6218, `caller_536300.dis`), scratch ≥52·N complex bytes:
  +0..8N A-accum, +8N..16N B-accum, +16N..24N z-powers, +24N..32N rotor-in (filled 1.0),
  +32N..36N phase ramp lVar1, +36N..44N rotor-out z1, +44N..52N lVar3(1.0)/lVar4(0.0)
 - `0x18052da00(lVar1, scale=2π/(os·sr), N)` = ramp fill; `0x18052dbc0(param_2, lVar4, N)` cplx-interleave
  → out slots get (cos,sin) = e^{+iθ}.
 - **Grid law**: `os=*(int*)(cfg+0x240080)=4` (mov DWORD PTR [rbx+0x240080],0x4 @0x18052ce3e),
  `sr=*(float*)(cfg+0x24)=44100.0` → generator `fs_total=os·sr=176400`; per-cell phase `2π/176400`;
  cell k ↔ freq k/os Hz (0.25 Hz steps). Phase-1 FS=44100 was a valid normalization (res depends on f/fc).
 - twin always = FUN_180535880 (type-info compare equal); sibling FUN_180536f90 dead.
 - case8: numBands `+0x814`=1, `+0x103`=1; `param_5=10^(sens_dB/20)`, `gain=sqrtf(param_5)=10^(sens_dB/40)`
  → fit gain=4.132 means stored `sens_dB@+0x101 ≈ 24.65`; freq=[band+0x804]=1000.0, Q=[band+0x80c].
 ## 5. LEVEL-WEIGHT FORMULA (Phase 4 — DECODED, NEGATIVE RESULT — NOTES_LEVEL.md §UPDATE 2026-08-18)
 Exact formula from decomp 22330 (FUN_180530d30):
 - `base=(2000/(sr·0.5))·iVar5/(bin+1)`, iVar5=NFFT/2+1
 - `w8=powf(base,0.25)`
 - `v=[0x540880]·0.25·w8·fVar12` (fVar12=4.0 if offline flag [0x5408b8], else 1.0)
 - `q=1/(1+v/([0x540880]·4))` (2nd pair 0x540884: q2=1/(1+v2/[0x540884]))
 - `dVar1=(sr/[0x1a0])·[0x1ac]·0.001`
 - `w=0.1^(1/(max(q·v,floor)·dVar1))`; bufs: 0x5406b8=w, 0x5406c8=1−w (and 0x5406d8/6e8 w/ 0x540884)
 - consts: c3d3c=0.25, c4334=4.0, c3d8c=0.5, c45b4=2000.0, c3f70=0.1 f32, c3e30=0.001 f64, c3f70=0.1
 - **Numerically**: w(500)=w(1000)=w(2000)≈0 (1e-20..1e-0.07), complements≈1 → but the weights are STILL
  APPLIED (as 1−w ≈ 1) by FUN_180529fe0 (`0x5407c8[band] += 0x5406c8·res_upper + 0x5406e8·res_lower`,
  thunk 003c40, kernel 0x1800752e0 = vfmadd213pd) — the form is weight·resonance, not dead code.
 - **TILT analysis (2026-08-18, checkpoint NEGATIVE result)**: warp `0x5406a8 = 0.87·x/(1+x/K)`,
  `K=exp(2.0723267)≈7.942`, `x=f/2000`, applied by FUN_180529fe0 as `mask *= 0x5406a8` (float, thunk 008700)
  BEFORE the FFT-conv. warp: 500→0.211, 1000→0.409, 2000→0.773 (lin ratio 2000/500=3.67).
  **BUT warp ≠ empirical tilt (ratio 1.27)**: `C(dB2000)=depth·warp(2000)·LUT_max = 0.864·0.772·0.667=0.445`,
  yet red2000=15.2 dB needs `C=0.826` ⇒ warp alone CANNOT produce deep-2000 (dual_b1q). Empirical tilt is the
  envelope of the BLEND term (0x5407c8 += weights·res + freq-axis 0x540698) + FFT-conv shaping, NOT warp —
  simple tilt→warp substitution in model_lut.py is INVALID. Nothing else decoded yet closes dual_b1q.
 - **FUN_180563ce0 decoded (2026-08-18)**: INIT of per-bin IIR level-trackers (0x156=342 bins, order=3.0f
  0x40400000, coeff A/B from 0x24c4780/4790, strided fill; third pass +0xffd0 fills 0x40e00000=7.0f).
  Not the update loop. Full disasm: `/tmp/opencode/f_563ce0.dis`.
 - **Level-path map (2026-08-18, NOTES_LEVEL.md)**: 0x563440 (LUT curve +0x188, gamma, 6 band-slots, combine
  →+0x2198, LUT pairs +0x98 at 0x3ff pts) + 0x56e3e0 twin-mask factory (N=1024, DOUBLE, LUT-ramp input) +
  0x563ce0 IIR init. 0x563440 NOT in decomp_funs.txt (search via /tmp/opencode/f_563440.dis + level_notes.md).
 - **DIAGNOSTIC BRIDGE DONE (2026-08-18) — model_fir.py**: реальная цепочка из FUN_180529fe0
  воспроизводит 36 точек до rmse=0.236 dB БЕЗ эмпирического tilt:
    `C(f) = g·LUT(log10(L0/res(f))) + w·warp(f)^a`, g=1.221, w=0.358, a=3.143.
  - dual500-константа = res_band(500;fc=500)=0.117 Q-НЕЗАВИСИМ, не tilt;
  - dual2000-глубина = АДДИТИВНЫЙ терм `w·warp^3.14` (на 1000 Гц вклад ≤0.021 — иначе t1k
    рушится; на 2000 = 0.149) — это аккумулятор `0x5407c8 += weights·res + mask` (шаг 5),
    НЕ мультипликация warp·LUT (та проваливается >10 dB);
  - warp^~3.14 ≈ π — подозрение на кратный каскад (0x540698 freq-axis, ∏0x540688, двойной FFT);
  - остаток 0.7 dB при Q=0.1 = форма LUT-колена 0.574, не закрывается per-bin → нужен FFT-уровень.
 - **FFT-CONV FULLY MAPPED (2026-08-18, NOTES_LEVEL.md §2026-08-18c, /tmp/opencode/f_52b570.dis)**:
  0x535a70 = dispatcher to base-CRT FFT (0x140a10/0x140a70); 0x540548/0x540550/0x540598 = FFT-plan
  tables (plan 0x540530, built by 0x52dc30), NOT windows. Per-band loop 0x52b550-0x52b8b5:
  mask→fwdFFT→invFFT→fill low/high (xmm13/xmm9 consts)→inv→cplx-op→fwd→`FIR *= 0x540658`→fill→inv→
  `FIR[0]=1,FIR[1]=0`→`*=0x540888`→copy out.
  **freq-axis 0x540698**: OFFLINE = constant scalar (NOT exp-formula — exp/rand = online dither).
  **Window 0x540658**: only ONE direct ref in dump (read at 0x52b771), written only indirectly —
  STATICALLY INVISIBLE → the only plausible source of warp^3.14/LUT-knee. Need runtime capture or
  full-pipeline empirical closure.
 ## 6. PHASE 5 PLAN (updated 2026-08-18 — bridge done, tilt replaced, FFT mined out)
 Decision (user): bridge FIRST. **DONE — bridge result above in §5.**
 Status: step 2 DONE, step 3 PARTIAL (static analysis exhausted).
 1. (15 min) Proof-of-state above; confirm twin + mdl files complete; repo at `60bf3a2`.
 2. **Replace empirical tilt** in model_lut.py with `g·LUT(xv) + w·warp(f)^a` — **DONE**, rmse 0.236 dB
   reproduced, committed `c8f97e4`. Optional refinement fit (2 LUT-knots) reaches 0.167 but is overfit.
 3. **Close the 0.7-dB gap**: FFT-conv loop mapped + freq-axis cleared (offline const) — **DONE/PARTIAL**,
   committed `60bf3a2`. Remaining unknown = 0x540658 window content (static-invisible). Options:
   (a) runtime capture of 0x540658, (b) skip and close empirically in full-pipeline step 5.
 4. Runtime values: per-bin IIR level trackers (0x563ce0), 0x540880/884, `[0x1a0]/[0x1ac]`, cell count N.
 5. Full pipeline + render diff vs `/home/m/soothe-bt/*.wav`: dB parity → byte parity.
 NEXT SESSION START (recommendation): skip deeper FFT mining; go straight to step 5 full pipeline
 (mask-chain already in model_lut.py B.12 at 0.236; freq-axis const; FFT-conv reduces to a per-bin
 smoothing + 0x540658 shaping that can be fit/absorbed). Runtime capture of 0x540658 only if step 5
 stalls on the Q=0.1 knee.
 ### STEP 5 STARTED (2026-08-18) — dB-parity harness DONE (commit f96bb42)
 - `/home/m/re-tools/render_parity.py`: measures steady-state per-tone reduction DIRECTLY on the
  reference wavs (Goertzel tone-amp over late steady window; handles 16/24-bit, mono/stereo).
  NOTE: Goertzel must use the SEQUENTIAL update form (s2=s1;s1=s0;s0=v+cw*s1-s2) — the tuple-unpack
  form is numerically unstable (overflows ~1500 steps).
 - Validation: measured dual Q-scan == stored dataset (dual.wav is 4s mono 16-bit; renders 6s stereo 24-bit).
 - **dB-PARITY RESULT (canonical B.12, 36 pts, real wavs): TOTAL rmse=0.268 dB**
  (dual500 0.130, dual2000 0.387, t1kq 0.320, t1k 0.076).
  residual structure: dual2000 -0.80@Q0.1..+0.36@Q10 (LUT-knee), t1kq +0.36..+0.26 (level region),
  t1k ±0.1 (good). Refit (g,w,a)=(1.224,0.379,3.405)→0.264; +2 LUT-knots→0.203 (plateau).
 - CONCLUSION: B.12 form plateau ~0.20-0.27 dB vs real renders; residual is structural (0x540658
  FFT shaping + level-curve region), matches plan expectation. Re-measured-on-wav numbers differ
  slightly from the stored model_lut.py dataset (t1kq ~+0.3 dB) — different steady-window method.
 - Open: full PIPELINE frame-render (STFT + per-frame mask + time constants) is the remaining big
  lift for byte-parity; dB-parity milestone achieved at 0.268 dB.
 Candidates that may consume Phase 5 time: source of sens≈24.65 (host ×2 over XML 12.0 — not found in dump),
 IAT\*0x181bab370 outside dump (non-blocking), final z¹ phase proof, the 0x563440 LUT curve (+0x188) values.
 ## 7. FILES LAYOUT
 - `/home/m/re-tools/dsp/` — twin.{hpp,cpp}, twin_check.cpp, CMakeLists.txt, build/
 - `/home/m/re-tools/model_lut.py` — model **B.12** (Q=xmlq, gain=10^(sens/20), depth 0.8639736175537109,
  `C=g·LUT(xv)+w·warp^a`, g=1.221/w=0.358/a=3.143) — 0.236 dB bridge model; PCHIP LUT nodes frozen
 - `/home/m/re-tools/model_fir.py` — bridge canonical source (rmse 0.236, committed b1066f3)
 - `/home/m/re-tools/framed_render.py` — full frame-render pilot (2026-08-19 params; modes dual|al)
 - `/home/m/re-tools/rwin_A0/A1/B0/C0.npy`, `r_freqaxis.npy`, `rwin_warp.npy` — live-window tables (48k)
 - `/home/m/re-tools/rtsnap.py` — live-process page snapshotter; snapshots /tmp/rt{A,B,C,D}.{raw,idx}
 - `/home/m/re-tools/handoff/decode_rpp_full.py`, `handoff/rpp_allparams.py` — RPP b64-XML full decoder
  (trim to len%4==0, `<?xml`@92, incl. nested processorStateData)
 - `/home/m/re-tools/soothe_mem.bin` — memory dump, offset=VA−0x180000000
 - `/home/m/re-tools/decomp_funs.txt` — decomp: FUN_180536300 (6218), FUN_180530d30 (22330),
  FUN_180533ec0 (123204), FUN_180529fe0 (missing body! vtable-only)
 - `/home/m/re-tools/handoff/` — THIS DOC, NOTES_TWIN.md, NOTES_LEVEL.md, phase1/*.py, nls_dasm/*.dis
 - `/home/m/soothe-bt/*.rpp` (decoder `/home/m/re-tools/dsp/...`; also `decode_rpp4.py` referenced) + `*.wav` refs
@@ -0,0 +1,150 @@
 ## Objective
 - Реверс-DSP oeksound soothe2 v1.1.2 VST3 → математическая модель (STFT, детектор резонансов, нотич-синтез). Текущая фаза — поведенческий параметр-свип через рендер Reaper RPP + статический декомпил (без декрипты бинарника пока).
 > **⚠️ ИСТОРИЧЕСКИЙ ДОКУМЕНТ (поведенческая модель v4, sim.py, ок. 2026-08-17)**.
 > Актуальный канон (bit-exact, P4) — `README.md`, `AGENTS.md`, `handoff/NOTES_LEVEL.md`.
 > Здесь — ранняя поведенческая фаза реверса, сохранена как справочник.
 ## Key Facts
 - Пользователь русскоязычный; sudo нет. Объект: `~/.wine/.../soothe2_x64.vst3` (PE32+, base 0x180000000), yabridge 5.1.1, Reaper 7.78.
 - Frida невозможна (Seccomp). Дамп памяти — `pread /proc/pid/mem` от родителя. pkill-ловушки: paths без `harneb`/`reaper` → `pkill -9 -x reaper`, `pkill -9 -f '[y]abridge'`.
 - Рендер-проект: `RENDER_RANGE 1 0 0 0 1000`, ITEM обязательно `POSITION 0`. Offline-рендер мгновенный → для живого дампа realtime (`RENDER_1X 1`) + `testtone.wav`.
 - VST-state = 92-байт JUCE-заголовок + base64 XML `<SOOTHE2STATE>` + 54 `<PARAM id=... value=.../>`; внутри — UI-state без DSP.
 ## Breakthrough (исправлено в этой сессии)
 - **Формат-адаптивность PARAM**: плагин применяет `<PARAM value>` ТОЛЬКО если строка сериализована в том же формате, что оригинал:
  - full-precision (depth, band freq/q...) → `%.16f` (напр. `"0.9000000000000000"`). Короткие строки → схлопывание (fallback).
  - short-формат `"X.0"` (selectivity, mix, mode, oversample, resolution) → ЛОМАЕТСЯ от `%.16f`.
  - `sweep.py`/`tt_sweep.py` адаптивны: >4 десятичных в оригинале → `%.16f`, 1-4 → то же число знаков, иначе — как есть.
 ## Синтез модели (поведенческая, v4 — реализована в sim.py, RMSE на settled ≤0.05dB по всем свипам)
 ```
 PARADИГМА (реализована):  out(t) = x(t) − amount_b(t)·bp_b(t; fc, Q)   [субтрактивный нотч, не микш-параллельный]
  amount(total) = A(depth)·S(sens)·H(sharp)·M(mode)      — LUT из измерений (пересечение, не аналит. фит)
  A(depth): 0.5/0.864/1/2/3/5/10/20 → 0.560/0.588/0.598/0.667/0.727/0.820/0.939/0.989
  S(sens):  0/6/12/24 → 0.48/0.76/1.0/1.0 (floor при sens=0!)
  H(sharp): 1/3/5/10 → 0.20/0.59/0.84/1.0
  M(mode):  0 → 0.745 (mode0 слабее), 1.. → 1.0
  Q(sel):   = 3.0+0.36·sel   (2nd-order bandpass; реал. даёт более крутые борта — приближ.)
  env(t):   one-pole, α_attack=1−e^(−1/(T_a·fs)), α_release=1−e^(−1/(T_r·fs))
            T_a = 0.02·e^(a/1.955) (0→20мс, 5→258мс, 10→3.3c);  T_r LUT из спадов (0→27мс, 5→140мс, 10→15c)
  детектор:  лауднеss полосы |bp| smoothed 5мс > thr 0.01 → цель amt, иначе 0   (порог между probe 0.0005 и burst 0.2)
 выход: x_out = mix/100·… — линейный кроссфейд: (100−mix)/100·x + mix/100·fully_reduced
       (mix=100 = полная обработка = ref; mix=0 → сухой; проверено на s_corr_mix: монот. линейно)
 ```
 Верификация (settled, RMSE по reduction dB на burst-окне):
 | параметр | реал | sim | RMSE | | параметр | реал | sim | RMSE |
 |---|---|---|--|---|---|---|---|---|
 | дефолт | 7.70 | 7.70 | 0.002 | sharp=1 | 1.11 | 1.11 | 0.002 |
 | depth=2 | 9.56 | 9.56 | 0.001 | sharp=3 | 3.72 | 3.71 | 0.006 |
 | depth=5 | 14.90 | 14.89 | 0.008 | attack=5 | 5.68 | 5.90 | 0.360 |
 | depth=10 | 24.27 | 24.25 | 0.018 | mode=0 | 5.03 | 5.00 | 0.023 |
 | depth=20 | 39.43 | 39.40 | 0.054 | mode=2 | 7.70 | 7.70 | 0.002 |
 | sens=0 | 2.89 | 2.89 | 0.001 | sens=24 | 7.70 | 7.70 | 0.002 |
 Q-профиль (probes-зонд): sim sel1 500→680@540→0.62@700 vs real 8.03→4.54→1.13; 2nd-order не покрывает
 крутые борта реала (реал резче ~×3): допустимо как 1-я итерация, RMSE макс по профилю ~1.6dB.
 Release-трассы: r0/r1/r2/r5 совпадают (sim против real 1.6s:3.5/4.3, 1.7s:1.5/1.9, 1.9s:0.3/0.3).
 ## Q/width-зонд (probes500/probes500b, drive 500Hz amp0.2 + справки 0.02)
 - Методика: длинный single-FFT на окне 1.7..2.9s (разрешение ~0.7Hz), справки на 500,505,510,520,540,
  555,575,600,700 (b-set) / 500,505,515,530,560,640,780 (a-set). Драйв-тон держит детект-клок,
  справки зондируют передаточную характеристику в установившемся режиме.
 - Данные (a-set) red@freq: sel1: 500:8.0 510:7.9 520:7.7 540:4.5 555:2.7 575:1.6 600:0.7 (плавный хвост)
  sel8: 500:8.7 510:9.1 520:6.6 540:1.3 555:0.1 575:0.0 600:0.0 (резкий срез) — Q уверенно растёт с sel.
 - Центраئة нотча на 500..510 чуть выше 500 (fs на 505..510) — трекинг слегка субадио.
 - Значения на дип-тонах 530/560/640 в b-наборе — из мусора (нет справки) и исключены; a-набор чистый.
 - ВАЖНО: dry-файлы дважды перезаписывались провальными dry-рендерами (RENDER_FILE не патчился по
  `.wav` напр. burst500L_byp → overwritten probes). ФИКС: патчить RENDER_FILE явно в sed до рендера.
 - data: оба svипа предсказуемо монотонны; таблица в summary выше.
 ## Верифицированные данные (все свипы с корректным форматом)
 - **depth (mir)**: монотонно distinct (vs-ref 496k→441k→...→347k@0.85→384k@1.0) — работает.
 - **selectivity**: 0/1/2/5 distinct (476k/466k/458k/461k vs ref); 10==20==50==100 (сатурация ≥10).
 - **oversample**: 0==ref; 1 distinct; 2==3==4 (сатурация на 2).
 - **resolution**: 0 distinct; 1==ref; 2 distinct (все distinct); 3==4.
 - **mix**: 0/25/50/75/100 всё distinct монотонно; 100==ref.
 - **mode**: 0 distinct; 1==2==3==4==ref.
 - **band1 freq**: tt_b1f_678 (дефолт banda freq) == tt_ref бит-в-бит; свип 100..8000 отрендерен.
 ## Прочие находки
 - 60-сек рендер p60_ref/p60_dep0/p60_dep1 подтвердил collapse класса при коротких строках.
 - rtdump.py живые дампы (регионы по VA, RA/WA) — DSP-кластер .data меняется между глубинами; раньше не совпали по layout (артефакт noise).
 ## Work State
 ### Completed
 - Свип-инфраструктура: sweep.py + tt_sweep.py (адаптив формата), рендер, сверочный анализ (spectrum.py медленный, нужно numpy).
 - Тест-тон: testtone.wav (2s, multi-tone 110Hz–14kHz с AM) для изоляции резонансов; tt_ref==tt_678 (0 diff).
 - Базовые свипы по всем основным параметрам.
 ### Active
 - **Извлечена статическая depth-кривая**: LUT 207 float @0x1826170e8..0x182617420, форма `0.302 + 0.698·sin(π/2·x)^0.94` (r²=0.99999), saved `depthcurve.npy`. ИДЕНТИЧНА у ref/dep0/dep1 → вшита в бинарь (build-time), не меняется от параметра = внутренний «depth→amount» маппинг.
 - **Адаптивная динамика (burst-тест, band1@500Hz, q≈1, depth 0.864)**: burst500.wav (2s: 0.5s тишины → 1.0s 500Hz burst amp 0.2 → 0.5s тишины, + 1000Hz amp 0.05 фон). `burst500_b1` (fx) vs `burst500_byp` (dry):
  - ATTACK: плавное включение нотч-ослабления, τ≈**18 мс** (фит на экспоненту A·(1−e^−t/τ), stable 17.6/17.7/18.8/18.8), установка −7.7dB бай-в-бай за ~100мс.
  - RELEASE: хвост после снятия бурста спадает за τ≈**5 мс** (после 1.5s fx-энергия 24→0 за ~30мс) — быстрый трелинг, симметрия фола не классическая (вероятно фильтр-транзиент, не медленный rel).
  - Глубина ослабления НЕ зависит от «порога» — это фикс. нотч на частоте band, amount задаётся curve(depth).
  - Рендерится через те же tt-файлы; РЕЗУЛЬТАТЫ в wav — 24-bit (`sampwidth=3`), читать через 24-bit decode (или любой np.frombuffer под sw), не 16-bit!
  - **τ-инвариантность** (burst-тест, band1@500Hz, win 10ms, Goertzel, фит к A(1−e^−t/τ)):
    - depth 0→0.86→1: A=6.39→7.70→7.91 dB; τ=18.3→16.8→16.5 ms (τ почти const) → depth контролирует amount, НЕ скорость.
    - selectivity: τ РАСТЁТ монотонно 12.3→16.8 ms при sel 0→10 (12.3/12.4/12.6/12.9/13.0/13.4/14.3/16.8) → Q/selectivity = time-const антреккинга.
    - depth-amount таблица (settled, band1@500): sel 0..6 монотонно −7.79→−9.08 dB, затем dip sel8=−8.79, sel10=−7.71 (non-monotonic у высоких sel, нюанс трекера).
    - «650Hz notch» в раннем FFT-анализе — АРТЕФАКТ деления на near-zero dry; реальные числа = ratio dry/fx только там, где dry>5% пика (pure 500Hz tone → окно вокруг 500).
 - **secret свипа: `patchparam.py`** обязателен для формат-сохранения: state-блок в RPP = base64, wrapped по 128-симв/строка с 2-sp indent и НОВОЙ строкой перед '>'. Ключевое: глубина пишется ТОЛЬКО `%.16f` (короткая `0.0` → collapse!), поэтому CLI `depth=0.0!` форсит полную точность.
 - Float-поиск по живому дампу: depth 301 hits (множество копий по bands), selectivity/sharpness → общий бакет (10.015, 9.997), mix → 4 копии 100.0, band freqs (678.76, 8242.67) → 0 точных (хранятся иначе, напр. Hz→binidx или через npf).
 - Живой diff dep0 vs dep1 в .data (0x182622130, 0x182622350, 0x1826223c0, 0x18262aca8, 0x182672fe4) — флоаты-состояние меняются; но регион содержит много мусора/nan (это .data буферы рендера, не коэфф).
 ### Провалы/уроки
 - ASLR между отдельными rtdump-запусками сдвигает раскладку регионов → прямого VA-сравнения НЕТ (dep0b 10 регионов, dep1b 9). Только те адреса, что реально в 0x18260a000-0x18267c000 (совпадают) сравнимы.
 ## Relevant Files
 - **`/home/m/re-tools/sim.py`** — поведенческий симулятор (LUT amount + 2nd-order bp Q(sel) + env attack/release + mix/mode). `simulate(x, fc, depth, sens, sharp, sel, mode, attack, release, mix)`. RMSE settled ≤0.05dB (см. таблицу выше).
 - **`/home/m/re-tools/verify_sim.py`** — RMSE-сверка симулятора против рендеров (mag@500 трасса). Использование: `verify_sim.py <fx.wav> --params depth=5 attack=5`.
 - **`/home/m/re-tools/fit_curves.py`** — фит LUT (остаётся как аналитический эквивалент; LUT в sim.py приоритетнее).
 - probes500/probes500b_wav+byp — зонды ширины нотча; prb_*/prb2_* — рендеры.
 - `synth_multi.py` (AM-комб 200–3000Hz резонансный), `burst500.wav` + `burst500_b1.rpp/.wav` + `burst500_byp.rpp/.wav` (burst-атак/релиз band1@500Hz), `burst_dep0/dep1/sel0/sel2/sel5.rpp/.wav` (τ-sweep), `comb.wav` + `comb_b1_*.rpp/wav`, `comb_dep_{0..1}`, `depthcurve.npy`. **Рендеры Reaper = 24-bit** — читать с `sampwidth=3`, не 16-bit!
 - `/home/m/re-tools/patchparam.py` — патчер PARAM в base64 state с формат-сохранением (128-wrap, `depth=...!` → %.16f).
 - `/home/m/re-tools/spec.py`, `probe.py`, `notch.py`, `synth.py`, `mkbase.py`, `tt_sweep.py`.
 - `/home/m/soothe-bt/tone1kq.wav` (чистый 1k, -18dBFS), `dual.wav` (500+2000Hz), `resonant.wav` (500Hz-резонанс).
 - `/home/m/soothe-bt/t1kq_*.rpp/wav`, `res_only1_*.rpp/wav`, `dual_b1_*.rpp/wav`.
 - `/home/m/re-tools/sweep.py` — адаптивный генератор RPP (tpl фиксирован, для смены файла использовать mkbase + tt_sweep).
 - `/home/m/soothe-bt/tt_base.rpp` (template: testtone + дефолтные парамы), `tt_ref.wav`.
 - `/home/m/soothe-bt/s_corr_{sel,os,res,mix,mode}*.wav` — верифицированные свипы на resonant.wav.
 - `/home/m/soothe-bt/render_v5.rpp`, `s_ref2.wav` (бит-в-бит оригинал).
 ## Фаза 3: Трекинг подтверждён (aug 16) — ARCHITECTURE REVISION
 - **Вход-зависимость**: ред@1k растёт с уровнем сигнала (band1=1000, тон 1k):
  `-27→-21→-15→-12→-9→-6→-3 dBFS` ⇒ `5.80→7.75→9.91→11.05→12.21→13.40→14.60 dB`. amount растёт с уровнем детектируемого резонанса (не бинарный порог).
 - **Трекинг нотча** (fresh renders `trk_b1_*`, вход tone1k 1k, band1∈{500,1000,2000}, длинная FFT): нотч ВСЕГДА на fc=тон=1k, band-частота НЕ двигает позицию режектора:
  - band=500 → 11.91dB, band=1000 → 15.82dB, band=2000 → 11.91dB, band1 off → 9.80dB, sens=0 → 9.80dB.
  - Единственный вклад band-EQ — ВЕС глубины детекции в области (совпадение band~тон даёт +4dB; off/sens0 −2dB).
 - **Мульти-пики**: dual (500+2000), оба нотча активны одновременно при sel=1..20 (~10dB оба) → детектор находит НЕСКОЛЬКО резонансов; selectivity НЕ регулирует число нотчей на явных тонах.
 - **Пользователь (иерархия)**: детектор оценивает каждую частоту спектра; band-EQ формирует его ВХОД (усиление/ослабление детекции по областям, у band2 sens может быть −12); selectivity = отбор пиков; sharpness = форма нотча; depth = глобальная глубина.
 - **Reaper-ловушка**: tt_sweep с ОТНОСИТЕЛЬНЫМ out_wav создаёт ПОДДИРЕКТОРИЮ → всегда абсолютный путь в RENDER_FILE.
 - Файлы: `trk_base.rpp`, `trk_b1_{500,1000,2000,eqoff,sens0}.rpp/.wav`, `sel_base.rpp`, `sel_{1,3,10,20}.rpp/.wav`, `lvl_{03..27}db.wav` + `lvl_t_*.rpp/.wav`.
 ### Фаза 3b: amount зависИТ от уровня сигнала (не просто порог)
 - lvl_sweep (тон 1k, band1@1000, depth=0.864): ред@1k = 5.80 / 7.75 / 9.91 / 11.05 / 12.21 / 13.40 / 14.60 для −27..−3 dBFS.
 - При depth=10: 21.67 / 24.15 / 26.63 / 27.86 / 29.09 / 30.33 / 31.56. Наклон d(ред)/d(level) ≈ 0.37–0.41 dB/dB обоих depth → amount ∝ уровень^p (монотонно с насыщением), НЕ бинарный порог. depth — масштаб, не floor.
 - Обе кривые монотонны и близки по форме (отношение ~2.2 при громком, ~3.7 при тихом → форма зависит от depth слабо).
 ### Фаза 4 (aug 16): КАРТА EQ-ВЕСОВ ДЕТЕКТОРА
 - **`on` полосы = enable детекторного веса**; off-полоса её sens НЕ отдаёт в детекцию. Подтверждения:
  - eqT_b1_b2/b3/b4/b2b3b4 и trk_b1_500 (sens off-полос любая) ВСЕ = 11.91dB (только от b1 sens12); eq_off_sensfac (все off, factory sens) = 9.80dB = нейтраль.
  - eq1778_b2s_12 (b2 on=0) = 11.81, b1on12_b2s12 (b2 on=1) = 17.60 — ЕДИНСТВЕННАЯ разница band2 on.
  - Баг в раннем clear_bands сбрасывал b1 sens→0 (eq_b1_on был sens0=нейтраль 9.80; правильный net_b1on12 sens12 = 11.91).
 - **S(sens) при совмещённом весе (band1@1k, тон 1k)**: sens 0/6/9/12/18/24 → boost над полом 0.00/3.16/4.65/6.02/6.02/6.02 dB.
  boost = 6.02·min(1, sens/12), насыщение +6dB при sens≥12 (не монотонно из-за сатурации детектора, не бага).
 - **W(f) — колокол EQ-веса** (тон 1k, одна полоса on sens12, варьирую band freq):
  - band1 (q≈1): 500→11.91(+2.11), 600→12.80(+3.00), 750→14.28(+4.48), 850→15.19(+5.39), 950→15.75(+5.95), 1000→15.82(+6.02), 1050→15.76, 1200→15.05(+5.25), 1500→13.44(+3.64), 2000→11.91(+2.11), 2500→11.16(+1.36). Симметричный широкий колокол (первый октав хвост +2.1dB, не спадает к 0).
  - band2 (q≈4.5): 600→10.15(+0.35), 750→10.78(+0.98), 850→11.97(+2.17), 1000→15.80(+6.00), 1200→11.68(+1.88), 1500→10.34(+0.54) — РЕЗКИЙ колокол. **Q полосы управляет шириной детекторного веса** (q=4.5 на −1 октаве теряет ~полностью, q=1 теряет только ~66%).
 - **Позиция нотча**: band freq не двигает режектор (всегда на резонансе тона); EQ-веса модулируют ТОЛЬКО глубину.
 - Новый модельный блок: `red(f_tone) = floor(level) + Σ_on boost_i`, `boost_i = 6.02·min(1,|sens_i|/12)·H_q(f_tone; fc_i, Q_i)`.
 - Файлы: `wf_{600,750,...,2500}.wav` (band1 q-sweep), `ws_{6,9,18,24}.wav` (aligned sens), `b2q_{600..1500}.wav` (band2 q=4.5 sweep), `b1on12_b2{on12,s12}.wav`.
 ### Фаза 5 (aug 16): STFT-параметры детектора (Stage 1)
 - **Хоп/задержка**: tone-jump зонд tj.wav (1k→1.5k в t=2.0s) → нотч пересаживается на новый резонанс за ~1 кадр ≈ 18мс (hop ≤ 512 сэмплов). Трекинг чирпа 400→2000Hz в реальном времени, dt≈0.03→0.05s (шум оценки FFT-бина, значимой задержки нет).
 - **Разрешение по частоте**: пары 1000+1040 → сливаются в один широкий нотч; 1000+1100 разделяются (центры ≈996/1104, но провал между ними всё ещё глубокий); 1000+1200 → два чётких отдельных нотча с полным восстановлением на 1100. → окно FFT ≈ 1024–2048 (бины 21–43Hz), multiple-нотч подтверждён.
 - **Энерготрекинг (fx@f по окну 2048, sync-Goertzel)**: до прыжка fx@1000 ≈ −24dB (ред ~14dB), после — fx@1500 ≈ −23.8dB. Двойной бурст (gap 10..100ms) → нотч на том же тоне не переоткрывается между бурстами (ред остаётся ~4dB floor) — bурст probe (осб) указывает на фиксированный window-центр, не пер-открытие.
 - Вывод для v5: STFT с окном ~2048, хоп ~512–1024; window function Hann; детектор по |X| per-bin → локальные пики (selectivity); EQ-веса W(f) как множитель sensitivity per-bin; per-пик нотч (bandstop Q=sharpness).
 - Файлы: `chirp_log_400_2000.wav`, `chirp_trk.rpp/.wav`, `pair_{1040,1100,1200}.wav` + `pair_trk_{40,100,200}`, `dburst_{10,20,50,100}.wav` + `dburst_trk_*`, `tj.wav` + `tj_trk.rpp/.wav`.
 ### Фаза 6 (aug 16): КОЛОКОЛ EQ-ВЕСА + SHARPNESS + level-floor (Stage 2 фит)
 - **H(f) — колокол**: `H(f) = 1/sqrt(1 + (Qeff·A)²)`, `A = f/fc − fc/f`. Qeff = **1.54·q^1.33** (q=1→1.54, 2→3.32, 3→5.64, 6→16.8). Воспроизводит кривую band1 (wf_*) с ошибкой ≤0.15dB по всему диапазону 500..2500.
 - **Sharpness → Q нотча**: psh_{1,3,5,10} (probes500, drive 500·0.2, band1@500): центр 505Hz red 1.3/4.3/6.8/8.7dB; ширина Q_notch≈1.15·sharp (sh10→~11, sh5→~6, sh3→~3, sh1→~1.2). Центр-глубина НЕ постоянна — растёт с sharp (1.3→8.7), т.е. sharp масштабирует amount И Q.
 - **level-floor (нейтраль)**: `neut_{24..3}` (все полосы off/sens0, тон 1k): red = 2.22/3.36/4.85/5.71/6.65/7.65 dB при −24/−18/−12/−9/−6/−3 dBFS rms. Монотонно, наклон ~0.31 dB/dB. `al_{*}` (b1@1000 sens12): 5.80/7.75/9.91/11.04/12.21/13.39 → boost = 3.59/4.39/5.06/5.33/5.56/5.74, **насыщается к ~6dB** с уровнем (не аддитивен в dB!).
 - Модель amount: `red(f) = curv(level·G_eq(f))` с насыщением (не линейный сдвиг в dB). depth масштабирует (фаза 3b). Для sim: 2D LUT level×W.
 - Файлы: `lvl_tone_lv{24,18,12,9,6,3}.wav`, `neut_{*}.rpp/.wav`, `al_{*}.rpp/.wav`, `psh_{1,3,5,10}.rpp/.wav`.
@@ -0,0 +1,30 @@
 import base64, re
 def decode(path):
    txt=open(path).read()
    lines=txt.split('\n')
    for i,l in enumerate(lines):
        if 'soothe2_x64.vst3' in l:
            b=[]
            j=i+1
            while j<len(lines):
                t=lines[j].strip()
                if t=='}': break
                b.append(t); j+=1
            raw=re.sub(r'\s','',''.join(b))
            raw=raw.rstrip('=')
            raw+='='*((-len(raw))%4)
            dec=base64.b64decode(raw)
            # dec is nested: outer base64 blob contains an inner chunked-xml whose first child is another base64 or actual xml
            # find '<PARAM ...' anywhere after utf8 decode of inner
            # Look for actual XML PARAM tags embedded (they are inside processorStateData attribute - base64 too)
            # Simplest: find 'band1 mode' bytes
            for pidname in [b'band1 mode', b'band1 q', b'band1 sens', b'band1 freq', b'band1 on']:
                i2=dec.find(pidname)
                if i2>=0:
                    seg=dec[i2-60:i2+80]
                    m=re.search(rb'<PARAM id="([^"]+)" value="([^"]+)"', seg)
                    if m:
                        print(f"  {m.group(1).decode():16s} = {m.group(2).decode()}")
            return
 for f in ['/home/m/soothe-bt/t1kq_b1f_800.rpp','/home/m/soothe-bt/dual_b1q_0.1.rpp','/home/m/soothe-bt/dual_b1q_10.0.rpp','/home/m/soothe-bt/dual_b1q_1.0.rpp','/home/m/soothe-bt/t1k_b1f_1100.rpp']:
    print("=== ",f.split('/')[-1]); decode(f)
@@ -0,0 +1,46 @@
 #!/usr/bin/env python3
 import base64, re, sys
 def full_decode(path):
    txt = open(path).read()
    lines = txt.split('\n')
    for i, l in enumerate(lines):
        if 'soothe2_x64.vst3' in l:
            b = []
            j = i + 1
            while j < len(lines):
                t = lines[j].strip()
                if t == '}':
                    break
                b.append(t)
                j += 1
            raw = re.sub(r'\s', '', ''.join(b))
            raw = raw.rstrip('=')
            raw += '=' * ((-len(raw)) % 4)
            dec = base64.b64decode(raw)
            # print first bytes overview
            print('decoded blob len', len(dec))
            # Look for inner base64 chunked xml (soothe2 vst uses nested b64)
            # search for PARAM tags in raw decoded bytes
            tags = re.findall(rb'<PARAM id="([^"]+)" value="([^"]+)"', dec)
            if tags:
                for pid, val in tags:
                    print(f'  {pid.decode():16s} = {val.decode()}')
            else:
                # try inner b64
                inner = re.sub(rb'[^A-Za-z0-9+/=]', b'', dec)
                inner = inner.rstrip(b'=')
                inner += b'=' * ((-len(inner)) % 4)
                try:
                    dec2 = base64.b64decode(inner)
                    tags = re.findall(rb'<PARAM id="([^"]+)" value="([^"]+)"', dec2)
                    for pid, val in tags:
                        print(f'  {pid.decode():16s} = {val.decode()}')
                except Exception as e:
                    print('  inner b64 fail:', e)
            return
    print('no soothe2_x64.vst3 found')
 for f in sys.argv[1:]:
    print('===', f.split('/')[-1])
    full_decode(f)
@@ -0,0 +1,55 @@
 #!/usr/bin/env python3
 """Emit dsp/leveltrack_data.hpp: live level-tracker A[] + mask scalars from
 handoff/rtctx_live.json (P1.5 realtime capture, render_long.rpp preset).
 """
 import json, os
 HERE = os.path.dirname(os.path.abspath(__file__))
 SRC = os.path.join(HERE, 'rtctx_live.json')
 OUT = os.path.join(HERE, '..', 'dsp', 'leveltrack_data.hpp')
 d = json.load(open(SRC))
 A_attack = d['A_4c0528']   # attack smoothing (fast), 341 double
 A_release = d['A_3c0510']  # release smoothing (slow), 341 double
 scalars = d['scalars']     # hex-offset -> float
 def fmt_doubles(a):
    return ", ".join("%.17g" % v for v in a)
 lines = []
 lines.append("// AUTOGENERATED from P1.5 realtime capture handoff/rtctx_live.json")
 lines.append("// (render_long.rpp: attack=0 release=0 selectivity=10 sharpness=10 depth=0.864).")
 lines.append("// Regenerate with handoff/emit_leveltrack.py. Do not edit by hand.")
 lines.append("#pragma once")
 lines.append("#include <cstddef>")
 lines.append("")
 lines.append("namespace ltk {")
 lines.append("")
 lines.append("// ctx = 0x2370040, marker +0x24 == 48000.0f (internal SR).")
 lines.append("constexpr float CTX_INTERNAL_SR = 48000.0f;")
 lines.append("")
 lines.append("// mask scalars (float) captured live:")
 for off, val in scalars.items():
    lines.append(f"constexpr float SCALAR{off.upper()} = {val}f;")
 lines.append("")
 lines.append(f"constexpr size_t LEVEL_NBINS = {len(A_attack)};")
 lines.append("")
 lines.append("// per-bin level-tracker IIR attack coefficients (0x4c0528):")
 lines.append(f"const double A_ATTACK[{len(A_attack)}] = {{")
 for i in range(0, len(A_attack), 6):
    lines.append("  " + ", ".join("%.17g" % v for v in A_attack[i:i+6]) + ",")
 lines.append("};")
 lines.append("")
 lines.append("// per-bin level-tracker IIR release coefficients (0x3c0510 == 0x2c04f8):")
 lines.append(f"const double A_RELEASE[{len(A_release)}] = {{")
 for i in range(0, len(A_release), 6):
    lines.append("  " + ", ".join("%.17g" % v for v in A_release[i:i+6]) + ",")
 lines.append("};")
 lines.append("")
 lines.append("}  // namespace ltk")
 open(OUT, 'w').write("\n".join(lines) + "\n")
 print(f"wrote {OUT}: attack={len(A_attack)} release={len(A_release)} scalars={len(scalars)}")
@@ -0,0 +1,70 @@
 #!/usr/bin/env python3
 """Emit dsp/tables_data.hpp: embed captured live .npy tables as C arrays.
 Sourced from handoff/rt*.npy (runtime capture of the DSP registry, NOTES_CAPTURE).
 Output is a single header so the port zero-copies the real tables the plugin used.
 """
 import numpy as np
 import os
 HERE = os.path.dirname(os.path.abspath(__file__))
 OUT = os.path.join(HERE, "..", "dsp", "tables_data.hpp")
 SPECS = [
    ("WIN_WINDOW", "rtwin_freq_44100.npy"),           # 0x540658 FFT-conv window, 8193 f32
    ("WIN_FREQAXIS", "rtfreqaxis_48000_internal.npy"),  # freq-axis 2048 f32 (internal SR 48000)
    ("WTA_WEIGHT", "rtwa_596.npy"),                    # [03] 0.596->0.126
    ("WTB_WEIGHT", "rtwb_404.npy"),                    # [04] 0.404->0.874
    ("WTC_WEIGHT", "rtwc_043.npy"),                    # [05]
    ("WTD_WEIGHT", "rtwd_956.npy"),                    # [06]
 ]
 WINDOW_N = 8193
 def fmt_floats(a):
    s = []
    for v in a:
        r = ("%.9g" % float(v)).encode().decode("ascii")
        s.append(r)
    return s
 def emit_header(out_path, blocks):
    with open(out_path, "w") as f:
        f.write("// AUTOGENERATED from runtime capture handoff/rt*.npy (NOTES_CAPTURE 2026-08-19).\n")
        f.write("// Do not edit by hand; regenerate with handoff/emit_tables.py.\n")
        f.write("#pragma once\n#include <cstddef>\n#include <cstdint>\n\n")
        for name, npy_file, arr in blocks:
            f.write(f"constexpr size_t {name}_COUNT = {arr.size};\n")
            f.write(f"const float {name}[{arr.size}] = {{\n")
            line = []
            for s in fmt_floats(arr):
                line.append(s)
                if len(line) == 8:
                    f.write("  " + ", ".join(line) + ",\n")
                    line = []
            if line:
                f.write("  " + ", ".join(line) + ",\n")
            f.write("};\n\n")
 def main():
    blocks = []
    for name, filename in SPECS:
        path = os.path.join(HERE, filename)
        if not os.path.exists(path):
            print(f"missing table {filename} — skipping")
            continue
        arr = np.load(path)
        arr = arr.ravel().astype(np.float32)
        if arr.shape[0] not in (WINDOW_N, 8193, 2048, 2049):
            print(f"unexpected size for {filename}: {arr.shape}")
        blocks.append((name, filename, arr))
        print(f"loaded {filename}: {arr.shape} {arr.dtype}")
    emit_header(OUT, blocks)
    print(f"wrote {OUT} ({sum(b[2].size for b in blocks)} floats)")
 if __name__ == "__main__":
    main()
@@ -0,0 +1,56 @@
 #!/usr/bin/env python3
 """extract_fft.py — extract FFT-related code regions from the rt snap into raw bins."""
 import struct, mmap, os
 SNAP = '/tmp/snap_rt.bin'
 OUT = '/tmp/fft/'
 # (name, addr, size)
 TARGETS = [
    ('cplx_mul_8440', 0x180008440, 0x80),
    ('cplx_mul_kernel_c440', 0x18000c440, 0x100),
    ('stage_bfc0', 0x18000bfc0, 0x600),
    ('stage_c5e0', 0x18000c5e0, 0x600),
    ('twiddle_loader_39b00', 0x180039b00, 0x400),
    ('plan_gen_2f980', 0x18002f980, 0x1200),
    ('dispatcher_535a70', 0x180535a70, 0x100),
    ('scalar_140a10', 0x180140a10, 0x200),
    ('vector_140a70', 0x180140a70, 0x200),
    ('fft_kernel_a5a0', 0x18001a5a0, 0x100),
 ]
 def main():
    os.makedirs(OUT, exist_ok=True)
    fd = os.open(SNAP, os.O_RDONLY)
    sz = os.fstat(fd).st_size
    mm = mmap.mmap(fd, 0, access=mmap.ACCESS_READ)
    # parse region index once
    regs = []
    i = 0
    while i + 16 <= sz:
        lo, n = struct.unpack_from('<QQ', mm, i)
        regs.append((lo, i + 16, n))  # (addr, data_off, size)
        i += 16 + n
    regs.sort()
    def readabs(addr, n):
        for lo, off, rn in regs:
            if lo <= addr < lo + rn and addr - lo + n <= rn:
                return mm[off + (addr - lo): off + (addr - lo) + n]
        return None
    for name, addr, n in TARGETS:
        b = readabs(addr, n)
        if b:
            with open(os.path.join(OUT, name + '.bin'), 'wb') as f:
                f.write(b)
            print('%-28s 0x%x %d bytes OK' % (name, addr, len(b)))
        else:
            print('%-28s 0x%x MISSING' % (name, addr))
    mm.close()
    os.close(fd)
 if __name__ == '__main__':
    main()
@@ -0,0 +1,648 @@
 /tmp/det/caller_536300.bin:     file format binary
 Disassembly of section .data:
 0000000180536300 <.data>:
   180536300:	48 89 5c 24 08       	mov    QWORD PTR [rsp+0x8],rbx
   180536305:	48 89 6c 24 18       	mov    QWORD PTR [rsp+0x18],rbp
   18053630a:	48 89 74 24 20       	mov    QWORD PTR [rsp+0x20],rsi
   18053630f:	48 89 54 24 10       	mov    QWORD PTR [rsp+0x10],rdx
   180536314:	57                   	push   rdi
   180536315:	41 54                	push   r12
   180536317:	41 55                	push   r13
   180536319:	41 56                	push   r14
   18053631b:	41 57                	push   r15
   18053631d:	48 83 ec 50          	sub    rsp,0x50
   180536321:	48 63 bc 24 a0 00 00 	movsxd rdi,DWORD PTR [rsp+0xa0]
   180536328:	00 
   180536329:	48 8d 15 88 aa 11 02 	lea    rdx,[rip+0x211aa88]        # 0x182650db8
   180536330:	4c 8b b4 24 b0 00 00 	mov    r14,QWORD PTR [rsp+0xb0]
   180536337:	00 
   180536338:	49 8b e9             	mov    rbp,r9
   18053633b:	0f 29 74 24 40       	movaps XMMWORD PTR [rsp+0x40],xmm6
   180536340:	49 8b d8             	mov    rbx,r8
   180536343:	66 0f 6e b1 80 00 24 	movd   xmm6,DWORD PTR [rcx+0x240080]
   18053634a:	00 
   18053634b:	8d 04 fd 00 00 00 00 	lea    eax,[rdi*8+0x0]
   180536352:	4c 63 d0             	movsxd r10,eax
   180536355:	8d 04 3f             	lea    eax,[rdi+rdi*1]
   180536358:	0f 5b f6             	cvtdq2ps xmm6,xmm6
   18053635b:	4f 8d 24 96          	lea    r12,[r14+r10*4]
   18053635f:	4c 63 d0             	movsxd r10,eax
   180536362:	4d 8d 3c bc          	lea    r15,[r12+rdi*4]
   180536366:	4c 89 a4 24 b0 00 00 	mov    QWORD PTR [rsp+0xb0],r12
   18053636d:	00 
   18053636e:	f3 0f 59 71 24       	mulss  xmm6,DWORD PTR [rcx+0x24]
   180536373:	48 8d 0d 3e aa 11 02 	lea    rcx,[rip+0x211aa3e]        # 0x182650db8
   18053637a:	4b 8d 34 97          	lea    rsi,[r15+r10*4]
   18053637e:	4e 8d 2c 96          	lea    r13,[rsi+r10*4]
   180536382:	ff 15 80 4c 67 01    	call   QWORD PTR [rip+0x1674c80]        # 0x181bab008
   180536388:	f2 0f 10 0d b8 de f8 	movsd  xmm1,QWORD PTR [rip+0x1f8deb8]        # 0x1824c4248
   18053638f:	01 
   180536390:	44 8b cf             	mov    r9d,edi
   180536393:	0f 5a c6             	cvtps2pd xmm0,xmm6
   180536396:	85 c0                	test   eax,eax
   180536398:	48 8b d3             	mov    rdx,rbx
   18053639b:	49 8b cc             	mov    rcx,r12
   18053639e:	0f 94 84 24 a8 00 00 	sete   BYTE PTR [rsp+0xa8]
   1805363a5:	00 
   1805363a6:	f2 0f 5e c8          	divsd  xmm1,xmm0
   1805363aa:	66 0f 5a d1          	cvtpd2ps xmm2,xmm1
   1805363ae:	e8 4d 76 ff ff       	call   0x18052da00
   1805363b3:	80 bd 10 08 00 00 00 	cmp    BYTE PTR [rbp+0x810],0x0
   1805363ba:	44 8b c7             	mov    r8d,edi
   1805363bd:	74 1a                	je     0x1805363d9
   1805363bf:	f3 0f 10 0d dd da f8 	movss  xmm1,DWORD PTR [rip+0x1f8dadd]        # 0x1824c3ea4
   1805363c6:	01 
   1805363c7:	48 8b 8c 24 88 00 00 	mov    rcx,QWORD PTR [rsp+0x88]
   1805363ce:	00 
   1805363cf:	e8 7c 77 ff ff       	call   0x18052db50
   1805363d4:	e9 88 01 00 00       	jmp    0x180536561
   1805363d9:	f3 0f 10 35 c3 da f8 	movss  xmm6,DWORD PTR [rip+0x1f8dac3]        # 0x1824c3ea4
   1805363e0:	01 
   1805363e1:	48 8b ce             	mov    rcx,rsi
   1805363e4:	0f 28 ce             	movaps xmm1,xmm6
   1805363e7:	e8 64 77 ff ff       	call   0x18052db50
   1805363ec:	44 8b c7             	mov    r8d,edi
   1805363ef:	0f 57 c9             	xorps  xmm1,xmm1
   1805363f2:	49 8b cd             	mov    rcx,r13
   1805363f5:	e8 56 77 ff ff       	call   0x18052db50
   1805363fa:	33 db                	xor    ebx,ebx
   1805363fc:	39 9d 14 08 00 00    	cmp    DWORD PTR [rbp+0x814],ebx
   180536402:	0f 8e 1e 01 00 00    	jle    0x180536526
   180536408:	0f 29 7c 24 30       	movaps XMMWORD PTR [rsp+0x30],xmm7
   18053640d:	4c 8d a5 18 08 00 00 	lea    r12,[rbp+0x818]
   180536414:	f2 0f 10 3d 24 dd f8 	movsd  xmm7,QWORD PTR [rip+0x1f8dd24]        # 0x1824c4140
   18053641b:	01 
   18053641c:	0f 1f 40 00          	nop    DWORD PTR [rax+0x0]
   180536420:	4c 8b 8c 24 b0 00 00 	mov    r9,QWORD PTR [rsp+0xb0]
   180536427:	00 
   180536428:	49 8b d6             	mov    rdx,r14
   18053642b:	4c 63 c3             	movsxd r8,ebx
   18053642e:	49 8b cf             	mov    rcx,r15
   180536431:	49 c1 e0 07          	shl    r8,0x7
   180536435:	4c 03 c5             	add    r8,rbp
   180536438:	89 7c 24 20          	mov    DWORD PTR [rsp+0x20],edi
   18053643c:	80 bc 24 a8 00 00 00 	cmp    BYTE PTR [rsp+0xa8],0x0
   180536443:	00 
   180536444:	74 07                	je     0x18053644d
   180536446:	e8 35 f4 ff ff       	call   0x180535880
   18053644b:	eb 05                	jmp    0x180536452
   18053644d:	e8 3e 0b 00 00       	call   0x180536f90
   180536452:	48 8d 15 5f a9 11 02 	lea    rdx,[rip+0x211a95f]        # 0x182650db8
   180536459:	48 8d 0d 58 a9 11 02 	lea    rcx,[rip+0x211a958]        # 0x182650db8
   180536460:	ff 15 a2 4b 67 01    	call   QWORD PTR [rip+0x1674ba2]        # 0x181bab008
   180536466:	44 8b c7             	mov    r8d,edi
   180536469:	49 8b d6             	mov    rdx,r14
   18053646c:	49 8b cf             	mov    rcx,r15
   18053646f:	85 c0                	test   eax,eax
   180536471:	75 07                	jne    0x18053647a
   180536473:	e8 68 b4 ac ff       	call   0x1800018e0
   180536478:	eb 05                	jmp    0x18053647f
   18053647a:	e8 51 b8 ac ff       	call   0x180001cd0
   18053647f:	48 8d 15 32 a9 11 02 	lea    rdx,[rip+0x211a932]        # 0x182650db8
   180536486:	48 8d 0d 2b a9 11 02 	lea    rcx,[rip+0x211a92b]        # 0x182650db8
   18053648d:	ff 15 75 4b 67 01    	call   QWORD PTR [rip+0x1674b75]        # 0x181bab008
   180536493:	44 8b c7             	mov    r8d,edi
   180536496:	48 8b d6             	mov    rdx,rsi
   180536499:	49 8b ce             	mov    rcx,r14
   18053649c:	85 c0                	test   eax,eax
   18053649e:	75 07                	jne    0x1805364a7
   1805364a0:	e8 5b bb ac ff       	call   0x180002000
   1805364a5:	eb 05                	jmp    0x1805364ac
   1805364a7:	e8 94 b7 ac ff       	call   0x180001c40
   1805364ac:	41 80 3c 24 00       	cmp    BYTE PTR [r12],0x0
   1805364b1:	74 5d                	je     0x180536510
   1805364b3:	48 8d 15 fe a8 11 02 	lea    rdx,[rip+0x211a8fe]        # 0x182650db8
   1805364ba:	48 8d 0d f7 a8 11 02 	lea    rcx,[rip+0x211a8f7]        # 0x182650db8
   1805364c1:	ff 15 41 4b 67 01    	call   QWORD PTR [rip+0x1674b41]        # 0x181bab008
   1805364c7:	44 8b c7             	mov    r8d,edi
   1805364ca:	49 8b d5             	mov    rdx,r13
   1805364cd:	48 8b ce             	mov    rcx,rsi
   1805364d0:	85 c0                	test   eax,eax
   1805364d2:	75 07                	jne    0x1805364db
   1805364d4:	e8 b7 bb ac ff       	call   0x180002090
   1805364d9:	eb 05                	jmp    0x1805364e0
   1805364db:	e8 d0 bc ac ff       	call   0x1800021b0
   1805364e0:	48 8d 15 d1 a8 11 02 	lea    rdx,[rip+0x211a8d1]        # 0x182650db8
   1805364e7:	48 8d 0d ca a8 11 02 	lea    rcx,[rip+0x211a8ca]        # 0x182650db8
   1805364ee:	ff 15 14 4b 67 01    	call   QWORD PTR [rip+0x1674b14]        # 0x181bab008
   1805364f4:	44 8b c7             	mov    r8d,edi
   1805364f7:	48 8b d6             	mov    rdx,rsi
   1805364fa:	85 c0                	test   eax,eax
   1805364fc:	75 0a                	jne    0x180536508
   1805364fe:	0f 28 c6             	movaps xmm0,xmm6
   180536501:	e8 9a b4 ac ff       	call   0x1800019a0
   180536506:	eb 08                	jmp    0x180536510
   180536508:	0f 28 c7             	movaps xmm0,xmm7
   18053650b:	e8 c0 bd ac ff       	call   0x1800022d0
   180536510:	ff c3                	inc    ebx
   180536512:	49 ff c4             	inc    r12
   180536515:	3b 9d 14 08 00 00    	cmp    ebx,DWORD PTR [rbp+0x814]
   18053651b:	0f 8c ff fe ff ff    	jl     0x180536420
   180536521:	0f 28 7c 24 30       	movaps xmm7,XMMWORD PTR [rsp+0x30]
   180536526:	8b 85 14 08 00 00    	mov    eax,DWORD PTR [rbp+0x814]
   18053652c:	83 f8 01             	cmp    eax,0x1
   18053652f:	7e 1d                	jle    0x18053654e
   180536531:	ff c8                	dec    eax
   180536533:	48 63 c8             	movsxd rcx,eax
   180536536:	80 bc 29 18 08 00 00 	cmp    BYTE PTR [rcx+rbp*1+0x818],0x0
   18053653d:	00 
   18053653e:	75 0e                	jne    0x18053654e
   180536540:	44 8b c7             	mov    r8d,edi
   180536543:	48 8b d6             	mov    rdx,rsi
   180536546:	49 8b cd             	mov    rcx,r13
   180536549:	e8 42 74 ff ff       	call   0x18052d990
   18053654e:	48 8b 8c 24 88 00 00 	mov    rcx,QWORD PTR [rsp+0x88]
   180536555:	00 
   180536556:	44 8b c7             	mov    r8d,edi
   180536559:	49 8b d5             	mov    rdx,r13
   18053655c:	e8 5f 76 ff ff       	call   0x18052dbc0
   180536561:	0f 28 74 24 40       	movaps xmm6,XMMWORD PTR [rsp+0x40]
   180536566:	4c 8d 5c 24 50       	lea    r11,[rsp+0x50]
   18053656b:	49 8b 5b 30          	mov    rbx,QWORD PTR [r11+0x30]
   18053656f:	49 8b 6b 40          	mov    rbp,QWORD PTR [r11+0x40]
   180536573:	49 8b 73 48          	mov    rsi,QWORD PTR [r11+0x48]
   180536577:	49 8b e3             	mov    rsp,r11
   18053657a:	41 5f                	pop    r15
   18053657c:	41 5e                	pop    r14
   18053657e:	41 5d                	pop    r13
   180536580:	41 5c                	pop    r12
   180536582:	5f                   	pop    rdi
   180536583:	c3                   	ret
   180536584:	cc                   	int3
   180536585:	cc                   	int3
   180536586:	cc                   	int3
   180536587:	cc                   	int3
   180536588:	cc                   	int3
   180536589:	cc                   	int3
   18053658a:	cc                   	int3
   18053658b:	cc                   	int3
   18053658c:	cc                   	int3
   18053658d:	cc                   	int3
   18053658e:	cc                   	int3
   18053658f:	cc                   	int3
   180536590:	48 8d 05 29 a7 11 02 	lea    rax,[rip+0x211a729]        # 0x182650cc0
   180536597:	c3                   	ret
   180536598:	cc                   	int3
   180536599:	cc                   	int3
   18053659a:	cc                   	int3
   18053659b:	cc                   	int3
   18053659c:	cc                   	int3
   18053659d:	cc                   	int3
   18053659e:	cc                   	int3
   18053659f:	cc                   	int3
   1805365a0:	48 8d 05 b1 5b f7 01 	lea    rax,[rip+0x1f75bb1]        # 0x1824ac158
   1805365a7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805365aa:	48 8b c2             	mov    rax,rdx
   1805365ad:	0f 10 41 08          	movups xmm0,XMMWORD PTR [rcx+0x8]
   1805365b1:	0f 11 42 08          	movups XMMWORD PTR [rdx+0x8],xmm0
   1805365b5:	c3                   	ret
   1805365b6:	cc                   	int3
   1805365b7:	cc                   	int3
   1805365b8:	cc                   	int3
   1805365b9:	cc                   	int3
   1805365ba:	cc                   	int3
   1805365bb:	cc                   	int3
   1805365bc:	cc                   	int3
   1805365bd:	cc                   	int3
   1805365be:	cc                   	int3
   1805365bf:	cc                   	int3
   1805365c0:	48 8d 05 d9 a9 11 02 	lea    rax,[rip+0x211a9d9]        # 0x182650fa0
   1805365c7:	c3                   	ret
   1805365c8:	cc                   	int3
   1805365c9:	cc                   	int3
   1805365ca:	cc                   	int3
   1805365cb:	cc                   	int3
   1805365cc:	cc                   	int3
   1805365cd:	cc                   	int3
   1805365ce:	cc                   	int3
   1805365cf:	cc                   	int3
   1805365d0:	48 8d 05 41 61 f7 01 	lea    rax,[rip+0x1f76141]        # 0x1824ac718
   1805365d7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805365da:	f2 0f 10 41 08       	movsd  xmm0,QWORD PTR [rcx+0x8]
   1805365df:	f2 0f 11 42 08       	movsd  QWORD PTR [rdx+0x8],xmm0
   1805365e4:	8b 41 10             	mov    eax,DWORD PTR [rcx+0x10]
   1805365e7:	89 42 10             	mov    DWORD PTR [rdx+0x10],eax
   1805365ea:	48 8b c2             	mov    rax,rdx
   1805365ed:	c3                   	ret
   1805365ee:	cc                   	int3
   1805365ef:	cc                   	int3
   1805365f0:	84 d2                	test   dl,dl
   1805365f2:	74 0a                	je     0x1805365fe
   1805365f4:	ba 20 00 00 00       	mov    edx,0x20
   1805365f9:	e9 f6 aa bf 00       	jmp    0x1811310f4
   1805365fe:	c3                   	ret
   1805365ff:	cc                   	int3
   180536600:	48 8d 05 d9 84 11 02 	lea    rax,[rip+0x21184d9]        # 0x18264eae0
   180536607:	c3                   	ret
   180536608:	cc                   	int3
   180536609:	cc                   	int3
   18053660a:	cc                   	int3
   18053660b:	cc                   	int3
   18053660c:	cc                   	int3
   18053660d:	cc                   	int3
   18053660e:	cc                   	int3
   18053660f:	cc                   	int3
   180536610:	48 8d 05 71 52 f7 01 	lea    rax,[rip+0x1f75271]        # 0x1824ab888
   180536617:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053661a:	0f 10 41 08          	movups xmm0,XMMWORD PTR [rcx+0x8]
   18053661e:	0f 11 42 08          	movups XMMWORD PTR [rdx+0x8],xmm0
   180536622:	48 8b 41 18          	mov    rax,QWORD PTR [rcx+0x18]
   180536626:	48 89 42 18          	mov    QWORD PTR [rdx+0x18],rax
   18053662a:	48 8b c2             	mov    rax,rdx
   18053662d:	c3                   	ret
   18053662e:	cc                   	int3
   18053662f:	cc                   	int3
   180536630:	48 8d 05 a9 a9 11 02 	lea    rax,[rip+0x211a9a9]        # 0x182650fe0
   180536637:	c3                   	ret
   180536638:	cc                   	int3
   180536639:	cc                   	int3
   18053663a:	cc                   	int3
   18053663b:	cc                   	int3
   18053663c:	cc                   	int3
   18053663d:	cc                   	int3
   18053663e:	cc                   	int3
   18053663f:	cc                   	int3
   180536640:	48 8d 05 a1 54 f7 01 	lea    rax,[rip+0x1f754a1]        # 0x1824abae8
   180536647:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053664a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053664e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536652:	48 8b c2             	mov    rax,rdx
   180536655:	c3                   	ret
   180536656:	cc                   	int3
   180536657:	cc                   	int3
   180536658:	cc                   	int3
   180536659:	cc                   	int3
   18053665a:	cc                   	int3
   18053665b:	cc                   	int3
   18053665c:	cc                   	int3
   18053665d:	cc                   	int3
   18053665e:	cc                   	int3
   18053665f:	cc                   	int3
   180536660:	48 8d 05 a9 a4 11 02 	lea    rax,[rip+0x211a4a9]        # 0x182650b10
   180536667:	c3                   	ret
   180536668:	cc                   	int3
   180536669:	cc                   	int3
   18053666a:	cc                   	int3
   18053666b:	cc                   	int3
   18053666c:	cc                   	int3
   18053666d:	cc                   	int3
   18053666e:	cc                   	int3
   18053666f:	cc                   	int3
   180536670:	48 8d 05 49 59 f7 01 	lea    rax,[rip+0x1f75949]        # 0x1824abfc0
   180536677:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053667a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053667e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536682:	48 8b c2             	mov    rax,rdx
   180536685:	c3                   	ret
   180536686:	cc                   	int3
   180536687:	cc                   	int3
   180536688:	cc                   	int3
   180536689:	cc                   	int3
   18053668a:	cc                   	int3
   18053668b:	cc                   	int3
   18053668c:	cc                   	int3
   18053668d:	cc                   	int3
   18053668e:	cc                   	int3
   18053668f:	cc                   	int3
   180536690:	48 8d 05 e9 a2 11 02 	lea    rax,[rip+0x211a2e9]        # 0x182650980
   180536697:	c3                   	ret
   180536698:	cc                   	int3
   180536699:	cc                   	int3
   18053669a:	cc                   	int3
   18053669b:	cc                   	int3
   18053669c:	cc                   	int3
   18053669d:	cc                   	int3
   18053669e:	cc                   	int3
   18053669f:	cc                   	int3
   1805366a0:	48 8d 05 79 48 f7 01 	lea    rax,[rip+0x1f74879]        # 0x1824aaf20
   1805366a7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805366aa:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   1805366ae:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   1805366b2:	48 8b c2             	mov    rax,rdx
   1805366b5:	c3                   	ret
   1805366b6:	cc                   	int3
   1805366b7:	cc                   	int3
   1805366b8:	cc                   	int3
   1805366b9:	cc                   	int3
   1805366ba:	cc                   	int3
   1805366bb:	cc                   	int3
   1805366bc:	cc                   	int3
   1805366bd:	cc                   	int3
   1805366be:	cc                   	int3
   1805366bf:	cc                   	int3
   1805366c0:	48 8d 05 b9 96 11 02 	lea    rax,[rip+0x21196b9]        # 0x18264fd80
   1805366c7:	c3                   	ret
   1805366c8:	cc                   	int3
   1805366c9:	cc                   	int3
   1805366ca:	cc                   	int3
   1805366cb:	cc                   	int3
   1805366cc:	cc                   	int3
   1805366cd:	cc                   	int3
   1805366ce:	cc                   	int3
   1805366cf:	cc                   	int3
   1805366d0:	48 8d 05 99 52 f7 01 	lea    rax,[rip+0x1f75299]        # 0x1824ab970
   1805366d7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805366da:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   1805366de:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   1805366e2:	48 8b c2             	mov    rax,rdx
   1805366e5:	c3                   	ret
   1805366e6:	cc                   	int3
   1805366e7:	cc                   	int3
   1805366e8:	cc                   	int3
   1805366e9:	cc                   	int3
   1805366ea:	cc                   	int3
   1805366eb:	cc                   	int3
   1805366ec:	cc                   	int3
   1805366ed:	cc                   	int3
   1805366ee:	cc                   	int3
   1805366ef:	cc                   	int3
   1805366f0:	48 8d 05 f1 9a 11 02 	lea    rax,[rip+0x2119af1]        # 0x1826501e8
   1805366f7:	c3                   	ret
   1805366f8:	cc                   	int3
   1805366f9:	cc                   	int3
   1805366fa:	cc                   	int3
   1805366fb:	cc                   	int3
   1805366fc:	cc                   	int3
   1805366fd:	cc                   	int3
   1805366fe:	cc                   	int3
   1805366ff:	cc                   	int3
   180536700:	48 8d 05 01 50 f7 01 	lea    rax,[rip+0x1f75001]        # 0x1824ab708
   180536707:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053670a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053670e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536712:	48 8b c2             	mov    rax,rdx
   180536715:	c3                   	ret
   180536716:	cc                   	int3
   180536717:	cc                   	int3
   180536718:	cc                   	int3
   180536719:	cc                   	int3
   18053671a:	cc                   	int3
   18053671b:	cc                   	int3
   18053671c:	cc                   	int3
   18053671d:	cc                   	int3
   18053671e:	cc                   	int3
   18053671f:	cc                   	int3
   180536720:	48 8d 05 d9 a5 11 02 	lea    rax,[rip+0x211a5d9]        # 0x182650d00
   180536727:	c3                   	ret
   180536728:	cc                   	int3
   180536729:	cc                   	int3
   18053672a:	cc                   	int3
   18053672b:	cc                   	int3
   18053672c:	cc                   	int3
   18053672d:	cc                   	int3
   18053672e:	cc                   	int3
   18053672f:	cc                   	int3
   180536730:	48 8d 05 31 48 f7 01 	lea    rax,[rip+0x1f74831]        # 0x1824aaf68
   180536737:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053673a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053673e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536742:	48 8b c2             	mov    rax,rdx
   180536745:	c3                   	ret
   180536746:	cc                   	int3
   180536747:	cc                   	int3
   180536748:	cc                   	int3
   180536749:	cc                   	int3
   18053674a:	cc                   	int3
   18053674b:	cc                   	int3
   18053674c:	cc                   	int3
   18053674d:	cc                   	int3
   18053674e:	cc                   	int3
   18053674f:	cc                   	int3
   180536750:	48 8d 05 49 9d 11 02 	lea    rax,[rip+0x2119d49]        # 0x1826504a0
   180536757:	c3                   	ret
   180536758:	cc                   	int3
   180536759:	cc                   	int3
   18053675a:	cc                   	int3
   18053675b:	cc                   	int3
   18053675c:	cc                   	int3
   18053675d:	cc                   	int3
   18053675e:	cc                   	int3
   18053675f:	cc                   	int3
   180536760:	48 8d 05 59 4e f7 01 	lea    rax,[rip+0x1f74e59]        # 0x1824ab5c0
   180536767:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053676a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053676e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536772:	48 8b c2             	mov    rax,rdx
   180536775:	c3                   	ret
   180536776:	cc                   	int3
   180536777:	cc                   	int3
   180536778:	cc                   	int3
   180536779:	cc                   	int3
   18053677a:	cc                   	int3
   18053677b:	cc                   	int3
   18053677c:	cc                   	int3
   18053677d:	cc                   	int3
   18053677e:	cc                   	int3
   18053677f:	cc                   	int3
   180536780:	48 8d 05 49 80 11 02 	lea    rax,[rip+0x2118049]        # 0x18264e7d0
   180536787:	c3                   	ret
   180536788:	cc                   	int3
   180536789:	cc                   	int3
   18053678a:	cc                   	int3
   18053678b:	cc                   	int3
   18053678c:	cc                   	int3
   18053678d:	cc                   	int3
   18053678e:	cc                   	int3
   18053678f:	cc                   	int3
   180536790:	48 8d 05 79 46 f7 01 	lea    rax,[rip+0x1f74679]        # 0x1824aae10
   180536797:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053679a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053679e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   1805367a2:	48 8b c2             	mov    rax,rdx
   1805367a5:	c3                   	ret
   1805367a6:	cc                   	int3
   1805367a7:	cc                   	int3
   1805367a8:	cc                   	int3
   1805367a9:	cc                   	int3
   1805367aa:	cc                   	int3
   1805367ab:	cc                   	int3
   1805367ac:	cc                   	int3
   1805367ad:	cc                   	int3
   1805367ae:	cc                   	int3
   1805367af:	cc                   	int3
   1805367b0:	48 8d 05 c9 a0 11 02 	lea    rax,[rip+0x211a0c9]        # 0x182650880
   1805367b7:	c3                   	ret
   1805367b8:	cc                   	int3
   1805367b9:	cc                   	int3
   1805367ba:	cc                   	int3
   1805367bb:	cc                   	int3
   1805367bc:	cc                   	int3
   1805367bd:	cc                   	int3
   1805367be:	cc                   	int3
   1805367bf:	cc                   	int3
   1805367c0:	48 8d 05 01 5c f7 01 	lea    rax,[rip+0x1f75c01]        # 0x1824ac3c8
   1805367c7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805367ca:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   1805367ce:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   1805367d2:	48 8b c2             	mov    rax,rdx
   1805367d5:	c3                   	ret
   1805367d6:	cc                   	int3
   1805367d7:	cc                   	int3
   1805367d8:	cc                   	int3
   1805367d9:	cc                   	int3
   1805367da:	cc                   	int3
   1805367db:	cc                   	int3
   1805367dc:	cc                   	int3
   1805367dd:	cc                   	int3
   1805367de:	cc                   	int3
   1805367df:	cc                   	int3
   1805367e0:	48 8d 05 29 80 11 02 	lea    rax,[rip+0x2118029]        # 0x18264e810
   1805367e7:	c3                   	ret
   1805367e8:	cc                   	int3
   1805367e9:	cc                   	int3
   1805367ea:	cc                   	int3
   1805367eb:	cc                   	int3
   1805367ec:	cc                   	int3
   1805367ed:	cc                   	int3
   1805367ee:	cc                   	int3
   1805367ef:	cc                   	int3
   1805367f0:	48 8d 05 39 51 f7 01 	lea    rax,[rip+0x1f75139]        # 0x1824ab930
   1805367f7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805367fa:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   1805367fe:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536802:	48 8b c2             	mov    rax,rdx
   180536805:	c3                   	ret
   180536806:	cc                   	int3
   180536807:	cc                   	int3
   180536808:	cc                   	int3
   180536809:	cc                   	int3
   18053680a:	cc                   	int3
   18053680b:	cc                   	int3
   18053680c:	cc                   	int3
   18053680d:	cc                   	int3
   18053680e:	cc                   	int3
   18053680f:	cc                   	int3
   180536810:	48 8d 05 59 8b 11 02 	lea    rax,[rip+0x2118b59]        # 0x18264f370
   180536817:	c3                   	ret
   180536818:	cc                   	int3
   180536819:	cc                   	int3
   18053681a:	cc                   	int3
   18053681b:	cc                   	int3
   18053681c:	cc                   	int3
   18053681d:	cc                   	int3
   18053681e:	cc                   	int3
   18053681f:	cc                   	int3
   180536820:	48 8d 05 81 48 f7 01 	lea    rax,[rip+0x1f74881]        # 0x1824ab0a8
   180536827:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053682a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053682e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536832:	48 8b c2             	mov    rax,rdx
   180536835:	c3                   	ret
   180536836:	cc                   	int3
   180536837:	cc                   	int3
   180536838:	cc                   	int3
   180536839:	cc                   	int3
   18053683a:	cc                   	int3
   18053683b:	cc                   	int3
   18053683c:	cc                   	int3
   18053683d:	cc                   	int3
   18053683e:	cc                   	int3
   18053683f:	cc                   	int3
   180536840:	48 8d 05 69 9e 11 02 	lea    rax,[rip+0x2119e69]        # 0x1826506b0
   180536847:	c3                   	ret
   180536848:	cc                   	int3
   180536849:	cc                   	int3
   18053684a:	cc                   	int3
   18053684b:	cc                   	int3
   18053684c:	cc                   	int3
   18053684d:	cc                   	int3
   18053684e:	cc                   	int3
   18053684f:	cc                   	int3
   180536850:	48 8d 05 c1 4f f7 01 	lea    rax,[rip+0x1f74fc1]        # 0x1824ab818
   180536857:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053685a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053685e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536862:	48 8b c2             	mov    rax,rdx
   180536865:	c3                   	ret
   180536866:	cc                   	int3
   180536867:	cc                   	int3
   180536868:	cc                   	int3
   180536869:	cc                   	int3
   18053686a:	cc                   	int3
   18053686b:	cc                   	int3
   18053686c:	cc                   	int3
   18053686d:	cc                   	int3
   18053686e:	cc                   	int3
   18053686f:	cc                   	int3
   180536870:	48 8d 05 e9 88 11 02 	lea    rax,[rip+0x21188e9]        # 0x18264f160
   180536877:	c3                   	ret
   180536878:	cc                   	int3
   180536879:	cc                   	int3
   18053687a:	cc                   	int3
   18053687b:	cc                   	int3
   18053687c:	cc                   	int3
   18053687d:	cc                   	int3
   18053687e:	cc                   	int3
   18053687f:	cc                   	int3
   180536880:	48 8d 05 19 59 f7 01 	lea    rax,[rip+0x1f75919]        # 0x1824ac1a0
   180536887:	48 89 02             	mov    QWORD PTR [rdx],rax
   18053688a:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   18053688e:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   180536892:	48 8b c2             	mov    rax,rdx
   180536895:	c3                   	ret
   180536896:	cc                   	int3
   180536897:	cc                   	int3
   180536898:	cc                   	int3
   180536899:	cc                   	int3
   18053689a:	cc                   	int3
   18053689b:	cc                   	int3
   18053689c:	cc                   	int3
   18053689d:	cc                   	int3
   18053689e:	cc                   	int3
   18053689f:	cc                   	int3
   1805368a0:	48 8d 05 d9 8d 11 02 	lea    rax,[rip+0x2118dd9]        # 0x18264f680
   1805368a7:	c3                   	ret
   1805368a8:	cc                   	int3
   1805368a9:	cc                   	int3
   1805368aa:	cc                   	int3
   1805368ab:	cc                   	int3
   1805368ac:	cc                   	int3
   1805368ad:	cc                   	int3
   1805368ae:	cc                   	int3
   1805368af:	cc                   	int3
   1805368b0:	48 8d 05 d9 49 f7 01 	lea    rax,[rip+0x1f749d9]        # 0x1824ab290
   1805368b7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805368ba:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   1805368be:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   1805368c2:	48 8b c2             	mov    rax,rdx
   1805368c5:	c3                   	ret
   1805368c6:	cc                   	int3
   1805368c7:	cc                   	int3
   1805368c8:	cc                   	int3
   1805368c9:	cc                   	int3
   1805368ca:	cc                   	int3
   1805368cb:	cc                   	int3
   1805368cc:	cc                   	int3
   1805368cd:	cc                   	int3
   1805368ce:	cc                   	int3
   1805368cf:	cc                   	int3
   1805368d0:	48 8d 05 29 8e 11 02 	lea    rax,[rip+0x2118e29]        # 0x18264f700
   1805368d7:	c3                   	ret
   1805368d8:	cc                   	int3
   1805368d9:	cc                   	int3
   1805368da:	cc                   	int3
   1805368db:	cc                   	int3
   1805368dc:	cc                   	int3
   1805368dd:	cc                   	int3
   1805368de:	cc                   	int3
   1805368df:	cc                   	int3
   1805368e0:	48 8d 05 81 4c f7 01 	lea    rax,[rip+0x1f74c81]        # 0x1824ab568
   1805368e7:	48 89 02             	mov    QWORD PTR [rdx],rax
   1805368ea:	48 8b 41 08          	mov    rax,QWORD PTR [rcx+0x8]
   1805368ee:	48 89 42 08          	mov    QWORD PTR [rdx+0x8],rax
   1805368f2:	48 8b c2             	mov    rax,rdx
   1805368f5:	c3                   	ret
   1805368f6:	cc                   	int3
   1805368f7:	cc                   	int3
   1805368f8:	cc                   	int3
   1805368f9:	cc                   	int3
   1805368fa:	cc                   	int3
   1805368fb:	cc                   	int3
   1805368fc:	cc                   	int3
   1805368fd:	cc                   	int3
   1805368fe:	cc                   	int3
   1805368ff:	cc                   	int3
@@ -0,0 +1,383 @@
 ; cons_52ec00 0x18052eb00-0x18052f100
 /tmp/slice.bin:     file format binary
 Disassembly of section .data:
 000000018052eb00 <.data>:
   18052eb00:	0f af c8             	imul   %eax,%ecx
   18052eb03:	8b 05 f7 cb 0f 02    	mov    0x20fcbf7(%rip),%eax        # 0x18262b700
   18052eb09:	44 0f af e0          	imul   %eax,%r12d
   18052eb0d:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052eb13:	05 e2 2c dc 06       	add    $0x6dc2ce2,%eax
   18052eb18:	44 2b e1             	sub    %ecx,%r12d
   18052eb1b:	25 7f 00 00 80       	and    $0x8000007f,%eax
   18052eb20:	7d 07                	jge    0x18052eb29
   18052eb22:	ff c8                	dec    %eax
   18052eb24:	83 c8 80             	or     $0xffffff80,%eax
   18052eb27:	ff c0                	inc    %eax
   18052eb29:	89 86 e0 04 24 00    	mov    %eax,0x2404e0(%rsi)
   18052eb2f:	48 63 c8             	movslq %eax,%rcx
   18052eb32:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052eb39:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052eb3e:	8b 05 c0 cb 0f 02    	mov    0x20fcbc0(%rip),%eax        # 0x18262b704
   18052eb44:	66 0f 6e d0          	movd   %eax,%xmm2
   18052eb48:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052eb4e:	ff c0                	inc    %eax
   18052eb50:	48 63 c8             	movslq %eax,%rcx
   18052eb53:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052eb5a:	0f 5b d2             	cvtdq2ps %xmm2,%xmm2
   18052eb5d:	f3 0f 59 d0          	mulss  %xmm0,%xmm2
   18052eb61:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052eb66:	8b 05 5c ca 0f 02    	mov    0x20fca5c(%rip),%eax        # 0x18262b5c8
   18052eb6c:	66 0f 6e c8          	movd   %eax,%xmm1
   18052eb70:	8b 05 8a cb 0f 02    	mov    0x20fcb8a(%rip),%eax        # 0x18262b700
   18052eb76:	0f 5b c9             	cvtdq2ps %xmm1,%xmm1
   18052eb79:	f3 0f 59 c8          	mulss  %xmm0,%xmm1
   18052eb7d:	f3 0f 59 ca          	mulss  %xmm2,%xmm1
   18052eb81:	f3 0f 58 cd          	addss  %xmm5,%xmm1
   18052eb85:	f3 0f 2c d9          	cvttss2si %xmm1,%ebx
   18052eb89:	0f af d8             	imul   %eax,%ebx
   18052eb8c:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052eb92:	05 9a f2 4b 00       	add    $0x4bf29a,%eax
   18052eb97:	89 9c 24 f8 00 00 00 	mov    %ebx,0xf8(%rsp)
   18052eb9e:	25 7f 00 00 80       	and    $0x8000007f,%eax
   18052eba3:	7d 07                	jge    0x18052ebac
   18052eba5:	ff c8                	dec    %eax
   18052eba7:	83 c8 80             	or     $0xffffff80,%eax
   18052ebaa:	ff c0                	inc    %eax
   18052ebac:	89 86 e0 04 24 00    	mov    %eax,0x2404e0(%rsi)
   18052ebb2:	48 63 c8             	movslq %eax,%rcx
   18052ebb5:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052ebbc:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052ebc1:	8b 05 3d cb 0f 02    	mov    0x20fcb3d(%rip),%eax        # 0x18262b704
   18052ebc7:	66 0f 6e d0          	movd   %eax,%xmm2
   18052ebcb:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052ebd1:	ff c0                	inc    %eax
   18052ebd3:	48 63 c8             	movslq %eax,%rcx
   18052ebd6:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052ebdd:	0f 5b d2             	cvtdq2ps %xmm2,%xmm2
   18052ebe0:	f3 0f 59 d0          	mulss  %xmm0,%xmm2
   18052ebe4:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052ebe9:	8b 05 d9 c9 0f 02    	mov    0x20fc9d9(%rip),%eax        # 0x18262b5c8
   18052ebef:	66 0f 6e c8          	movd   %eax,%xmm1
   18052ebf3:	8b 05 07 cb 0f 02    	mov    0x20fcb07(%rip),%eax        # 0x18262b700
   18052ebf9:	0f 5b c9             	cvtdq2ps %xmm1,%xmm1
   18052ebfc:	f3 0f 59 c8          	mulss  %xmm0,%xmm1
   18052ec00:	f3 0f 59 ca          	mulss  %xmm2,%xmm1
   18052ec04:	f3 0f 58 cd          	addss  %xmm5,%xmm1
   18052ec08:	f3 0f 2c c9          	cvttss2si %xmm1,%ecx
   18052ec0c:	0f af c8             	imul   %eax,%ecx
   18052ec0f:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052ec15:	05 94 62 cb 00       	add    $0xcb6294,%eax
   18052ec1a:	66 0f 6e f1          	movd   %ecx,%xmm6
   18052ec1e:	0f 5b f6             	cvtdq2ps %xmm6,%xmm6
   18052ec21:	25 7f 00 00 80       	and    $0x8000007f,%eax
   18052ec26:	7d 07                	jge    0x18052ec2f
   18052ec28:	ff c8                	dec    %eax
   18052ec2a:	83 c8 80             	or     $0xffffff80,%eax
   18052ec2d:	ff c0                	inc    %eax
   18052ec2f:	89 86 e0 04 24 00    	mov    %eax,0x2404e0(%rsi)
   18052ec35:	48 63 c8             	movslq %eax,%rcx
   18052ec38:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052ec3f:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052ec44:	8b 05 ba ca 0f 02    	mov    0x20fcaba(%rip),%eax        # 0x18262b704
   18052ec4a:	66 0f 6e e0          	movd   %eax,%xmm4
   18052ec4e:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052ec54:	ff c0                	inc    %eax
   18052ec56:	48 63 c8             	movslq %eax,%rcx
   18052ec59:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052ec60:	0f 5b e4             	cvtdq2ps %xmm4,%xmm4
   18052ec63:	f3 0f 59 e0          	mulss  %xmm0,%xmm4
   18052ec67:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052ec6c:	8b 05 56 c9 0f 02    	mov    0x20fc956(%rip),%eax        # 0x18262b5c8
   18052ec72:	66 0f 6e d8          	movd   %eax,%xmm3
   18052ec76:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052ec7c:	0f 5b db             	cvtdq2ps %xmm3,%xmm3
   18052ec7f:	05 e0 b5 10 00       	add    $0x10b5e0,%eax
   18052ec84:	f3 0f 59 d8          	mulss  %xmm0,%xmm3
   18052ec88:	25 7f 00 00 80       	and    $0x8000007f,%eax
   18052ec8d:	7d 07                	jge    0x18052ec96
   18052ec8f:	ff c8                	dec    %eax
   18052ec91:	83 c8 80             	or     $0xffffff80,%eax
   18052ec94:	ff c0                	inc    %eax
   18052ec96:	f3 0f 10 3d 06 52 f9 	movss  0x1f95206(%rip),%xmm7        # 0x1824c3ea4
   18052ec9d:	01 
   18052ec9e:	33 ff                	xor    %edi,%edi
   18052eca0:	89 86 e0 04 24 00    	mov    %eax,0x2404e0(%rsi)
   18052eca6:	48 63 c8             	movslq %eax,%rcx
   18052eca9:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052ecb0:	f3 0f 59 dc          	mulss  %xmm4,%xmm3
   18052ecb4:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052ecb9:	8b 05 45 ca 0f 02    	mov    0x20fca45(%rip),%eax        # 0x18262b704
   18052ecbf:	f3 0f 58 dd          	addss  %xmm5,%xmm3
   18052ecc3:	66 0f 6e d0          	movd   %eax,%xmm2
   18052ecc7:	8b 86 e0 04 24 00    	mov    0x2404e0(%rsi),%eax
   18052eccd:	ff c0                	inc    %eax
   18052eccf:	48 63 c8             	movslq %eax,%rcx
   18052ecd2:	48 8b 86 b0 08 54 00 	mov    0x5408b0(%rsi),%rax
   18052ecd9:	0f 5b d2             	cvtdq2ps %xmm2,%xmm2
   18052ecdc:	f3 44 0f 2c eb       	cvttss2si %xmm3,%r13d
   18052ece1:	f3 0f 59 d0          	mulss  %xmm0,%xmm2
   18052ece5:	f3 0f 10 04 88       	movss  (%rax,%rcx,4),%xmm0
   18052ecea:	8b 05 d8 c8 0f 02    	mov    0x20fc8d8(%rip),%eax        # 0x18262b5c8
   18052ecf0:	66 0f 6e c8          	movd   %eax,%xmm1
   18052ecf4:	8b 05 06 ca 0f 02    	mov    0x20fca06(%rip),%eax        # 0x18262b700
   18052ecfa:	0f 5b c9             	cvtdq2ps %xmm1,%xmm1
   18052ecfd:	f3 0f 59 c8          	mulss  %xmm0,%xmm1
   18052ed01:	f3 0f 59 ca          	mulss  %xmm2,%xmm1
   18052ed05:	f3 0f 58 cd          	addss  %xmm5,%xmm1
   18052ed09:	f3 0f 2c c9          	cvttss2si %xmm1,%ecx
   18052ed0d:	0f af c8             	imul   %eax,%ecx
   18052ed10:	8b 05 ea c9 0f 02    	mov    0x20fc9ea(%rip),%eax        # 0x18262b700
   18052ed16:	44 0f af e8          	imul   %eax,%r13d
   18052ed1a:	44 03 e9             	add    %ecx,%r13d
   18052ed1d:	84 d2                	test   %dl,%dl
   18052ed1f:	0f 85 95 03 00 00    	jne    0x18052f0ba
   18052ed25:	48 8b ce             	mov    %rsi,%rcx
   18052ed28:	e8 d3 e0 ff ff       	call   0x18052ce00
   18052ed2d:	8b 86 a0 01 00 00    	mov    0x1a0(%rsi),%eax
   18052ed33:	48 8d 8e 68 06 54 00 	lea    0x540668(%rsi),%rcx
   18052ed3a:	0f af c3             	imul   %ebx,%eax
   18052ed3d:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052ed40:	c1 e0 03             	shl    $0x3,%eax
   18052ed43:	89 86 68 08 54 00    	mov    %eax,0x540868(%rsi)
   18052ed49:	99                   	cltd
   18052ed4a:	2b c2                	sub    %edx,%eax
   18052ed4c:	d1 f8                	sar    $1,%eax
   18052ed4e:	8d 14 18             	lea    (%rax,%rbx,1),%edx
   18052ed51:	89 96 6c 08 54 00    	mov    %edx,0x54086c(%rsi)
   18052ed57:	03 d2                	add    %edx,%edx
   18052ed59:	e8 32 f4 ff ff       	call   0x18052e190
   18052ed5e:	8b 96 6c 08 54 00    	mov    0x54086c(%rsi),%edx
   18052ed64:	48 8d 8e 98 06 54 00 	lea    0x540698(%rsi),%rcx
   18052ed6b:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052ed6e:	e8 1d f4 ff ff       	call   0x18052e190
   18052ed73:	8b d7                	mov    %edi,%edx
   18052ed75:	39 be a0 06 54 00    	cmp    %edi,0x5406a0(%rsi)
   18052ed7b:	7e 1e                	jle    0x18052ed9b
   18052ed7d:	8b cf                	mov    %edi,%ecx
   18052ed7f:	90                   	nop
   18052ed80:	48 8b 86 98 06 54 00 	mov    0x540698(%rsi),%rax
   18052ed87:	48 8d 49 04          	lea    0x4(%rcx),%rcx
   18052ed8b:	ff c2                	inc    %edx
   18052ed8d:	f3 0f 11 74 01 fc    	movss  %xmm6,-0x4(%rcx,%rax,1)
   18052ed93:	3b 96 a0 06 54 00    	cmp    0x5406a0(%rsi),%edx
   18052ed99:	7c e5                	jl     0x18052ed80
   18052ed9b:	8b 96 6c 08 54 00    	mov    0x54086c(%rsi),%edx
   18052eda1:	48 8d 8e a8 06 54 00 	lea    0x5406a8(%rsi),%rcx
   18052eda8:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052edab:	e8 e0 f3 ff ff       	call   0x18052e190
   18052edb0:	8b 96 6c 08 54 00    	mov    0x54086c(%rsi),%edx
   18052edb6:	48 8d 8e b8 06 54 00 	lea    0x5406b8(%rsi),%rcx
   18052edbd:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052edc0:	e8 cb f3 ff ff       	call   0x18052e190
   18052edc5:	8b 96 6c 08 54 00    	mov    0x54086c(%rsi),%edx
   18052edcb:	48 8d 8e c8 06 54 00 	lea    0x5406c8(%rsi),%rcx
   18052edd2:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052edd5:	e8 b6 f3 ff ff       	call   0x18052e190
   18052edda:	8b 96 6c 08 54 00    	mov    0x54086c(%rsi),%edx
   18052ede0:	48 8d 8e d8 06 54 00 	lea    0x5406d8(%rsi),%rcx
   18052ede7:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052edea:	e8 a1 f3 ff ff       	call   0x18052e190
   18052edef:	8b 96 6c 08 54 00    	mov    0x54086c(%rsi),%edx
   18052edf5:	48 8d 8e e8 06 54 00 	lea    0x5406e8(%rsi),%rcx
   18052edfc:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052edff:	e8 8c f3 ff ff       	call   0x18052e190
   18052ee04:	45 8b fc             	mov    %r12d,%r15d
   18052ee07:	44 3b 66 30          	cmp    0x30(%rsi),%r12d
   18052ee0b:	0f 8d ee 01 00 00    	jge    0x18052efff
   18052ee11:	45 8b f5             	mov    %r13d,%r14d
   18052ee14:	49 63 dc             	movslq %r12d,%rbx
   18052ee17:	45 0f af f5          	imul   %r13d,%r14d
   18052ee1b:	48 81 c3 73 40 05 00 	add    $0x54073,%rbx
   18052ee22:	48 c1 e3 04          	shl    $0x4,%rbx
   18052ee26:	41 c1 e6 0e          	shl    $0xe,%r14d
   18052ee2a:	48 03 de             	add    %rsi,%rbx
   18052ee2d:	0f 1f 00             	nopl   (%rax)
   18052ee30:	8b ae 6c 08 54 00    	mov    0x54086c(%rsi),%ebp
   18052ee36:	8b 83 50 ff ff ff    	mov    -0xb0(%rbx),%eax
   18052ee3c:	3b c5                	cmp    %ebp,%eax
   18052ee3e:	74 3e                	je     0x18052ee7e
   18052ee40:	85 c0                	test   %eax,%eax
   18052ee42:	7e 11                	jle    0x18052ee55
   18052ee44:	48 8b 8b 48 ff ff ff 	mov    -0xb8(%rbx),%rcx
   18052ee4b:	48 85 c9             	test   %rcx,%rcx
   18052ee4e:	74 05                	je     0x18052ee55
   18052ee50:	e8 6b 22 ad ff       	call   0x1800010c0
   18052ee55:	85 ed                	test   %ebp,%ebp
   18052ee57:	48 89 bb 48 ff ff ff 	mov    %rdi,-0xb8(%rbx)
   18052ee5e:	0f 48 ef             	cmovs  %edi,%ebp
   18052ee61:	89 ab 50 ff ff ff    	mov    %ebp,-0xb0(%rbx)
   18052ee67:	85 ed                	test   %ebp,%ebp
   18052ee69:	74 3e                	je     0x18052eea9
   18052ee6b:	8d 0c ad 00 00 00 00 	lea    0x0(,%rbp,4),%ecx
   18052ee72:	e8 09 22 ad ff       	call   0x180001080
   18052ee77:	48 89 83 48 ff ff ff 	mov    %rax,-0xb8(%rbx)
   18052ee7e:	8b d7                	mov    %edi,%edx
   18052ee80:	39 bb 50 ff ff ff    	cmp    %edi,-0xb0(%rbx)
   18052ee86:	7e 21                	jle    0x18052eea9
   18052ee88:	48 8b cf             	mov    %rdi,%rcx
   18052ee8b:	0f 1f 44 00 00       	nopl   0x0(%rax,%rax,1)
   18052ee90:	48 8b 83 48 ff ff ff 	mov    -0xb8(%rbx),%rax
   18052ee97:	48 8d 49 04          	lea    0x4(%rcx),%rcx
   18052ee9b:	ff c2                	inc    %edx
   18052ee9d:	89 7c 01 fc          	mov    %edi,-0x4(%rcx,%rax,1)
   18052eea1:	3b 93 50 ff ff ff    	cmp    -0xb0(%rbx),%edx
   18052eea7:	7c e7                	jl     0x18052ee90
   18052eea9:	8b 03                	mov    (%rbx),%eax
   18052eeab:	41 3b c6             	cmp    %r14d,%eax
   18052eeae:	74 35                	je     0x18052eee5
   18052eeb0:	85 c0                	test   %eax,%eax
   18052eeb2:	7e 0e                	jle    0x18052eec2
   18052eeb4:	48 8b 4b f8          	mov    -0x8(%rbx),%rcx
   18052eeb8:	48 85 c9             	test   %rcx,%rcx
   18052eebb:	74 05                	je     0x18052eec2
   18052eebd:	e8 fe 21 ad ff       	call   0x1800010c0
   18052eec2:	45 85 f6             	test   %r14d,%r14d
   18052eec5:	48 89 7b f8          	mov    %rdi,-0x8(%rbx)
   18052eec9:	41 8b ce             	mov    %r14d,%ecx
   18052eecc:	0f 48 cf             	cmovs  %edi,%ecx
   18052eecf:	89 0b                	mov    %ecx,(%rbx)
   18052eed1:	85 c9                	test   %ecx,%ecx
   18052eed3:	74 31                	je     0x18052ef06
   18052eed5:	8d 0c 8d 00 00 00 00 	lea    0x0(,%rcx,4),%ecx
   18052eedc:	e8 9f 21 ad ff       	call   0x180001080
   18052eee1:	48 89 43 f8          	mov    %rax,-0x8(%rbx)
   18052eee5:	8b d7                	mov    %edi,%edx
   18052eee7:	39 3b                	cmp    %edi,(%rbx)
   18052eee9:	7e 1b                	jle    0x18052ef06
   18052eeeb:	48 8b cf             	mov    %rdi,%rcx
   18052eeee:	66 90                	xchg   %ax,%ax
   18052eef0:	48 8b 43 f8          	mov    -0x8(%rbx),%rax
   18052eef4:	48 8d 49 04          	lea    0x4(%rcx),%rcx
   18052eef8:	ff c2                	inc    %edx
   18052eefa:	c7 44 01 fc 00 00 80 	movl   $0x3f800000,-0x4(%rcx,%rax,1)
   18052ef01:	3f 
   18052ef02:	3b 13                	cmp    (%rbx),%edx
   18052ef04:	7c ea                	jl     0x18052eef0
   18052ef06:	8b ae 6c 08 54 00    	mov    0x54086c(%rsi),%ebp
   18052ef0c:	8b 83 80 00 00 00    	mov    0x80(%rbx),%eax
   18052ef12:	3b c5                	cmp    %ebp,%eax
   18052ef14:	74 35                	je     0x18052ef4b
   18052ef16:	85 c0                	test   %eax,%eax
   18052ef18:	7e 0e                	jle    0x18052ef28
   18052ef1a:	48 8b 4b 78          	mov    0x78(%rbx),%rcx
   18052ef1e:	48 85 c9             	test   %rcx,%rcx
   18052ef21:	74 05                	je     0x18052ef28
   18052ef23:	e8 98 21 ad ff       	call   0x1800010c0
   18052ef28:	85 ed                	test   %ebp,%ebp
   18052ef2a:	48 89 7b 78          	mov    %rdi,0x78(%rbx)
   18052ef2e:	0f 48 ef             	cmovs  %edi,%ebp
   18052ef31:	89 ab 80 00 00 00    	mov    %ebp,0x80(%rbx)
   18052ef37:	85 ed                	test   %ebp,%ebp
   18052ef39:	74 3b                	je     0x18052ef76
   18052ef3b:	8d 0c ad 00 00 00 00 	lea    0x0(,%rbp,4),%ecx
   18052ef42:	e8 39 21 ad ff       	call   0x180001080
   18052ef47:	48 89 43 78          	mov    %rax,0x78(%rbx)
   18052ef4b:	8b d7                	mov    %edi,%edx
   18052ef4d:	39 bb 80 00 00 00    	cmp    %edi,0x80(%rbx)
   18052ef53:	7e 21                	jle    0x18052ef76
   18052ef55:	48 8b cf             	mov    %rdi,%rcx
   18052ef58:	0f 1f 84 00 00 00 00 	nopl   0x0(%rax,%rax,1)
   18052ef5f:	00 
   18052ef60:	48 8b 43 78          	mov    0x78(%rbx),%rax
   18052ef64:	48 8d 49 04          	lea    0x4(%rcx),%rcx
   18052ef68:	ff c2                	inc    %edx
   18052ef6a:	89 7c 01 fc          	mov    %edi,-0x4(%rcx,%rax,1)
   18052ef6e:	3b 93 80 00 00 00    	cmp    0x80(%rbx),%edx
   18052ef74:	7c ea                	jl     0x18052ef60
   18052ef76:	8b ae 6c 08 54 00    	mov    0x54086c(%rsi),%ebp
   18052ef7c:	8b 83 a0 00 00 00    	mov    0xa0(%rbx),%eax
   18052ef82:	3b c5                	cmp    %ebp,%eax
   18052ef84:	74 3e                	je     0x18052efc4
   18052ef86:	85 c0                	test   %eax,%eax
   18052ef88:	7e 11                	jle    0x18052ef9b
   18052ef8a:	48 8b 8b 98 00 00 00 	mov    0x98(%rbx),%rcx
   18052ef91:	48 85 c9             	test   %rcx,%rcx
   18052ef94:	74 05                	je     0x18052ef9b
   18052ef96:	e8 25 21 ad ff       	call   0x1800010c0
   18052ef9b:	85 ed                	test   %ebp,%ebp
   18052ef9d:	48 89 bb 98 00 00 00 	mov    %rdi,0x98(%rbx)
   18052efa4:	0f 48 ef             	cmovs  %edi,%ebp
   18052efa7:	89 ab a0 00 00 00    	mov    %ebp,0xa0(%rbx)
   18052efad:	85 ed                	test   %ebp,%ebp
   18052efaf:	74 3d                	je     0x18052efee
   18052efb1:	8d 0c ad 00 00 00 00 	lea    0x0(,%rbp,4),%ecx
   18052efb8:	e8 c3 20 ad ff       	call   0x180001080
   18052efbd:	48 89 83 98 00 00 00 	mov    %rax,0x98(%rbx)
   18052efc4:	8b d7                	mov    %edi,%edx
   18052efc6:	39 bb a0 00 00 00    	cmp    %edi,0xa0(%rbx)
   18052efcc:	7e 20                	jle    0x18052efee
   18052efce:	48 8b cf             	mov    %rdi,%rcx
   18052efd1:	48 8b 83 98 00 00 00 	mov    0x98(%rbx),%rax
   18052efd8:	48 8d 49 04          	lea    0x4(%rcx),%rcx
   18052efdc:	ff c2                	inc    %edx
   18052efde:	c7 44 01 fc 00 00 80 	movl   $0x3f800000,-0x4(%rcx,%rax,1)
   18052efe5:	3f 
   18052efe6:	3b 93 a0 00 00 00    	cmp    0xa0(%rbx),%edx
   18052efec:	7c e3                	jl     0x18052efd1
   18052efee:	41 ff c7             	inc    %r15d
   18052eff1:	48 83 c3 10          	add    $0x10,%rbx
   18052eff5:	44 3b 7e 30          	cmp    0x30(%rsi),%r15d
   18052eff9:	0f 8c 31 fe ff ff    	jl     0x18052ee30
   18052efff:	8b 96 68 08 54 00    	mov    0x540868(%rsi),%edx
   18052f005:	48 8d 8e f8 06 54 00 	lea    0x5406f8(%rsi),%rcx
   18052f00c:	c1 e2 03             	shl    $0x3,%edx
   18052f00f:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052f012:	e8 79 f1 ff ff       	call   0x18052e190
   18052f017:	8b 96 68 08 54 00    	mov    0x540868(%rsi),%edx
   18052f01d:	48 8d 8e 08 07 54 00 	lea    0x540708(%rsi),%rcx
   18052f024:	41 0f af d5          	imul   %r13d,%edx
   18052f028:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052f02b:	e8 60 f1 ff ff       	call   0x18052e190
   18052f030:	8b 96 68 08 54 00    	mov    0x540868(%rsi),%edx
   18052f036:	48 8d 8e 18 07 54 00 	lea    0x540718(%rsi),%rcx
   18052f03d:	41 0f af d5          	imul   %r13d,%edx
   18052f041:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052f044:	e8 47 f1 ff ff       	call   0x18052e190
   18052f049:	8b 96 68 08 54 00    	mov    0x540868(%rsi),%edx
   18052f04f:	48 8d 8e 98 07 54 00 	lea    0x540798(%rsi),%rcx
   18052f056:	0f 57 d2             	xorps  %xmm2,%xmm2
   18052f059:	e8 32 f1 ff ff       	call   0x18052e190
   18052f05e:	44 8b c7             	mov    %edi,%r8d
   18052f061:	39 be a0 07 54 00    	cmp    %edi,0x5407a0(%rsi)
   18052f067:	7e 26                	jle    0x18052f08f
   18052f069:	48 8b cf             	mov    %rdi,%rcx
   18052f06c:	0f 1f 40 00          	nopl   0x0(%rax)
   18052f070:	48 8b 86 98 07 54 00 	mov    0x540798(%rsi),%rax
   18052f077:	48 8d 49 04          	lea    0x4(%rcx),%rcx
   18052f07b:	41 ff c0             	inc    %r8d
   18052f07e:	c7 44 01 fc 00 00 80 	movl   $0x3f800000,-0x4(%rcx,%rax,1)
   18052f085:	3f 
   18052f086:	44 3b 86 a0 07 54 00 	cmp    0x5407a0(%rsi),%r8d
   18052f08d:	7c e1                	jl     0x18052f070
   18052f08f:	44 8b 86 68 08 54 00 	mov    0x540868(%rsi),%r8d
   18052f096:	48 8d 8e 30 05 54 00 	lea    0x540530(%rsi),%rcx
   18052f09d:	8b 56 64             	mov    0x64(%rsi),%edx
   18052f0a0:	e8 8b eb ff ff       	call   0x18052dc30
   18052f0a5:	0f 28 df             	movaps %xmm7,%xmm3
   18052f0a8:	48 8d 8e 30 05 54 00 	lea    0x540530(%rsi),%rcx
   18052f0af:	45 8b c5             	mov    %r13d,%r8d
   18052f0b2:	41 8b d5             	mov    %r13d,%edx
   18052f0b5:	e8 46 0d 00 00       	call   0x18052fe00
   18052f0ba:	45 8b cc             	mov    %r12d,%r9d
   18052f0bd:	4c 8b ac 24 00 01 00 	mov    0x100(%rsp),%r13
   18052f0c4:	00 
   18052f0c5:	44 3b 66 30          	cmp    0x30(%rsi),%r12d
   18052f0c9:	0f 8d a8 00 00 00    	jge    0x18052f177
   18052f0cf:	49 63 cc             	movslq %r12d,%rcx
   18052f0d2:	48 81 c1 7b 40 05 00 	add    $0x5407b,%rcx
   18052f0d9:	48 c1 e1 04          	shl    $0x4,%rcx
   18052f0dd:	48 03 ce             	add    %rsi,%rcx
   18052f0e0:	44 8b c7             	mov    %edi,%r8d
   18052f0e3:	39 79 80             	cmp    %edi,-0x80(%rcx)
   18052f0e6:	7e 22                	jle    0x18052f10a
   18052f0e8:	48 8b d7             	mov    %rdi,%rdx
   18052f0eb:	0f 1f 44 00 00       	nopl   0x0(%rax,%rax,1)
   18052f0f0:	48 8b 81 78 ff ff ff 	mov    -0x88(%rcx),%rax
   18052f0f7:	48 8d 52 04          	lea    0x4(%rdx),%rdx
   18052f0fb:	41 ff c0             	inc    %r8d
   18052f0fe:	f3                   	repz
   18052f0ff:	0f                   	.byte 0xf
@@ -0,0 +1,275 @@
 ; cons_536100 0x180536000-0x180536400
 /tmp/slice.bin:     file format binary
 Disassembly of section .data:
 0000000180536000 <.data>:
   180536000:	e8 bb b0 ac ff       	call   0x1800010c0
   180536005:	90                   	nop
   180536006:	48 89 bb 58 07 54 00 	mov    %rdi,0x540758(%rbx)
   18053600d:	89 bb 60 07 54 00    	mov    %edi,0x540760(%rbx)
   180536013:	39 bb 50 07 54 00    	cmp    %edi,0x540750(%rbx)
   180536019:	7e 12                	jle    0x18053602d
   18053601b:	48 8b 8b 48 07 54 00 	mov    0x540748(%rbx),%rcx
   180536022:	48 85 c9             	test   %rcx,%rcx
   180536025:	74 06                	je     0x18053602d
   180536027:	e8 94 b0 ac ff       	call   0x1800010c0
   18053602c:	90                   	nop
   18053602d:	48 89 bb 48 07 54 00 	mov    %rdi,0x540748(%rbx)
   180536034:	89 bb 50 07 54 00    	mov    %edi,0x540750(%rbx)
   18053603a:	48 8d 8b 28 07 54 00 	lea    0x540728(%rbx),%rcx
   180536041:	4c 8d 0d d8 81 ff ff 	lea    -0x7e28(%rip),%r9        # 0x18052e220
   180536048:	ba 10 00 00 00       	mov    $0x10,%edx
   18053604d:	44 8d 42 f2          	lea    -0xe(%rdx),%r8d
   180536051:	e8 c6 b6 bf 00       	call   0x18113171c
   180536056:	90                   	nop
   180536057:	39 bb 20 07 54 00    	cmp    %edi,0x540720(%rbx)
   18053605d:	7e 12                	jle    0x180536071
   18053605f:	48 8b 8b 18 07 54 00 	mov    0x540718(%rbx),%rcx
   180536066:	48 85 c9             	test   %rcx,%rcx
   180536069:	74 06                	je     0x180536071
   18053606b:	e8 50 b0 ac ff       	call   0x1800010c0
   180536070:	90                   	nop
   180536071:	48 89 bb 18 07 54 00 	mov    %rdi,0x540718(%rbx)
   180536078:	89 bb 20 07 54 00    	mov    %edi,0x540720(%rbx)
   18053607e:	39 bb 10 07 54 00    	cmp    %edi,0x540710(%rbx)
   180536084:	7e 12                	jle    0x180536098
   180536086:	48 8b 8b 08 07 54 00 	mov    0x540708(%rbx),%rcx
   18053608d:	48 85 c9             	test   %rcx,%rcx
   180536090:	74 06                	je     0x180536098
   180536092:	e8 29 b0 ac ff       	call   0x1800010c0
   180536097:	90                   	nop
   180536098:	48 89 bb 08 07 54 00 	mov    %rdi,0x540708(%rbx)
   18053609f:	89 bb 10 07 54 00    	mov    %edi,0x540710(%rbx)
   1805360a5:	39 bb 00 07 54 00    	cmp    %edi,0x540700(%rbx)
   1805360ab:	7e 12                	jle    0x1805360bf
   1805360ad:	48 8b 8b f8 06 54 00 	mov    0x5406f8(%rbx),%rcx
   1805360b4:	48 85 c9             	test   %rcx,%rcx
   1805360b7:	74 06                	je     0x1805360bf
   1805360b9:	e8 02 b0 ac ff       	call   0x1800010c0
   1805360be:	90                   	nop
   1805360bf:	48 89 bb f8 06 54 00 	mov    %rdi,0x5406f8(%rbx)
   1805360c6:	89 bb 00 07 54 00    	mov    %edi,0x540700(%rbx)
   1805360cc:	39 bb f0 06 54 00    	cmp    %edi,0x5406f0(%rbx)
   1805360d2:	7e 12                	jle    0x1805360e6
   1805360d4:	48 8b 8b e8 06 54 00 	mov    0x5406e8(%rbx),%rcx
   1805360db:	48 85 c9             	test   %rcx,%rcx
   1805360de:	74 06                	je     0x1805360e6
   1805360e0:	e8 db af ac ff       	call   0x1800010c0
   1805360e5:	90                   	nop
   1805360e6:	48 89 bb e8 06 54 00 	mov    %rdi,0x5406e8(%rbx)
   1805360ed:	89 bb f0 06 54 00    	mov    %edi,0x5406f0(%rbx)
   1805360f3:	39 bb e0 06 54 00    	cmp    %edi,0x5406e0(%rbx)
   1805360f9:	7e 12                	jle    0x18053610d
   1805360fb:	48 8b 8b d8 06 54 00 	mov    0x5406d8(%rbx),%rcx
   180536102:	48 85 c9             	test   %rcx,%rcx
   180536105:	74 06                	je     0x18053610d
   180536107:	e8 b4 af ac ff       	call   0x1800010c0
   18053610c:	90                   	nop
   18053610d:	48 89 bb d8 06 54 00 	mov    %rdi,0x5406d8(%rbx)
   180536114:	89 bb e0 06 54 00    	mov    %edi,0x5406e0(%rbx)
   18053611a:	39 bb d0 06 54 00    	cmp    %edi,0x5406d0(%rbx)
   180536120:	7e 12                	jle    0x180536134
   180536122:	48 8b 8b c8 06 54 00 	mov    0x5406c8(%rbx),%rcx
   180536129:	48 85 c9             	test   %rcx,%rcx
   18053612c:	74 06                	je     0x180536134
   18053612e:	e8 8d af ac ff       	call   0x1800010c0
   180536133:	90                   	nop
   180536134:	48 89 bb c8 06 54 00 	mov    %rdi,0x5406c8(%rbx)
   18053613b:	89 bb d0 06 54 00    	mov    %edi,0x5406d0(%rbx)
   180536141:	39 bb c0 06 54 00    	cmp    %edi,0x5406c0(%rbx)
   180536147:	7e 12                	jle    0x18053615b
   180536149:	48 8b 8b b8 06 54 00 	mov    0x5406b8(%rbx),%rcx
   180536150:	48 85 c9             	test   %rcx,%rcx
   180536153:	74 06                	je     0x18053615b
   180536155:	e8 66 af ac ff       	call   0x1800010c0
   18053615a:	90                   	nop
   18053615b:	48 89 bb b8 06 54 00 	mov    %rdi,0x5406b8(%rbx)
   180536162:	89 bb c0 06 54 00    	mov    %edi,0x5406c0(%rbx)
   180536168:	39 bb b0 06 54 00    	cmp    %edi,0x5406b0(%rbx)
   18053616e:	7e 12                	jle    0x180536182
   180536170:	48 8b 8b a8 06 54 00 	mov    0x5406a8(%rbx),%rcx
   180536177:	48 85 c9             	test   %rcx,%rcx
   18053617a:	74 06                	je     0x180536182
   18053617c:	e8 3f af ac ff       	call   0x1800010c0
   180536181:	90                   	nop
   180536182:	48 89 bb a8 06 54 00 	mov    %rdi,0x5406a8(%rbx)
   180536189:	89 bb b0 06 54 00    	mov    %edi,0x5406b0(%rbx)
   18053618f:	39 bb a0 06 54 00    	cmp    %edi,0x5406a0(%rbx)
   180536195:	7e 12                	jle    0x1805361a9
   180536197:	48 8b 8b 98 06 54 00 	mov    0x540698(%rbx),%rcx
   18053619e:	48 85 c9             	test   %rcx,%rcx
   1805361a1:	74 06                	je     0x1805361a9
   1805361a3:	e8 18 af ac ff       	call   0x1800010c0
   1805361a8:	90                   	nop
   1805361a9:	48 89 bb 98 06 54 00 	mov    %rdi,0x540698(%rbx)
   1805361b0:	89 bb a0 06 54 00    	mov    %edi,0x5406a0(%rbx)
   1805361b6:	48 8d 8b 78 06 54 00 	lea    0x540678(%rbx),%rcx
   1805361bd:	4c 8d 0d 5c 80 ff ff 	lea    -0x7fa4(%rip),%r9        # 0x18052e220
   1805361c4:	ba 10 00 00 00       	mov    $0x10,%edx
   1805361c9:	44 8d 42 f2          	lea    -0xe(%rdx),%r8d
   1805361cd:	e8 4a b5 bf 00       	call   0x18113171c
   1805361d2:	90                   	nop
   1805361d3:	39 bb 70 06 54 00    	cmp    %edi,0x540670(%rbx)
   1805361d9:	7e 12                	jle    0x1805361ed
   1805361db:	48 8b 8b 68 06 54 00 	mov    0x540668(%rbx),%rcx
   1805361e2:	48 85 c9             	test   %rcx,%rcx
   1805361e5:	74 06                	je     0x1805361ed
   1805361e7:	e8 d4 ae ac ff       	call   0x1800010c0
   1805361ec:	90                   	nop
   1805361ed:	48 89 bb 68 06 54 00 	mov    %rdi,0x540668(%rbx)
   1805361f4:	89 bb 70 06 54 00    	mov    %edi,0x540670(%rbx)
   1805361fa:	48 8d 8b 30 05 54 00 	lea    0x540530(%rbx),%rcx
   180536201:	e8 2a 7b ff ff       	call   0x18052dd30
   180536206:	90                   	nop
   180536207:	89 bb cc 04 24 00    	mov    %edi,0x2404cc(%rbx)
   18053620d:	48 8b 8b c0 04 24 00 	mov    0x2404c0(%rbx),%rcx
   180536214:	ff 15 8e 50 67 01    	call   *0x167508e(%rip)        # 0x181bab2a8
   18053621a:	90                   	nop
   18053621b:	48 8d 8b d8 03 00 00 	lea    0x3d8(%rbx),%rcx
   180536222:	e8 59 70 ff ff       	call   0x18052d280
   180536227:	90                   	nop
   180536228:	48 8d 05 49 5e f7 01 	lea    0x1f75e49(%rip),%rax        # 0x1824ac078
   18053622f:	48 89 83 d0 03 00 00 	mov    %rax,0x3d0(%rbx)
   180536236:	48 8b cb             	mov    %rbx,%rcx
   180536239:	48 8b 5c 24 40       	mov    0x40(%rsp),%rbx
   18053623e:	48 83 c4 30          	add    $0x30,%rsp
   180536242:	5f                   	pop    %rdi
   180536243:	e9 58 38 ff ff       	jmp    0x180529aa0
   180536248:	cc                   	int3
   180536249:	cc                   	int3
   18053624a:	cc                   	int3
   18053624b:	cc                   	int3
   18053624c:	cc                   	int3
   18053624d:	cc                   	int3
   18053624e:	cc                   	int3
   18053624f:	cc                   	int3
   180536250:	f2 0f 10 0d e8 de f8 	movsd  0x1f8dee8(%rip),%xmm1        # 0x1824c4140
   180536257:	01 
   180536258:	48 8d 81 10 00 08 00 	lea    0x80010(%rcx),%rax
   18053625f:	f2 0f 11 4c 24 08    	movsd  %xmm1,0x8(%rsp)
   180536265:	ba 00 80 00 00       	mov    $0x8000,%edx
   18053626a:	f2 0f 10 44 24 08    	movsd  0x8(%rsp),%xmm0
   180536270:	4c 8b c1             	mov    %rcx,%r8
   180536273:	c6 41 04 00          	movb   $0x0,0x4(%rcx)
   180536277:	48 c7 81 10 00 10 00 	movq   $0x0,0x100010(%rcx)
   18053627e:	00 00 00 00 
   180536282:	8b ca                	mov    %edx,%ecx
   180536284:	66 0f c6 c0 00       	shufpd $0x0,%xmm0,%xmm0
   180536289:	a8 0f                	test   $0xf,%al
   18053628b:	75 13                	jne    0x1805362a0
   18053628d:	0f 1f 00             	nopl   (%rax)
   180536290:	0f 11 00             	movups %xmm0,(%rax)
   180536293:	48 8d 40 10          	lea    0x10(%rax),%rax
   180536297:	48 83 e9 01          	sub    $0x1,%rcx
   18053629b:	75 f3                	jne    0x180536290
   18053629d:	eb 0e                	jmp    0x1805362ad
   18053629f:	90                   	nop
   1805362a0:	0f 11 00             	movups %xmm0,(%rax)
   1805362a3:	48 8d 40 10          	lea    0x10(%rax),%rax
   1805362a7:	48 83 e9 01          	sub    $0x1,%rcx
   1805362ab:	75 f3                	jne    0x1805362a0
   1805362ad:	f2 0f 11 4c 24 08    	movsd  %xmm1,0x8(%rsp)
   1805362b3:	49 8d 40 10          	lea    0x10(%r8),%rax
   1805362b7:	f2 0f 10 44 24 08    	movsd  0x8(%rsp),%xmm0
   1805362bd:	66 0f c6 c0 00       	shufpd $0x0,%xmm0,%xmm0
   1805362c2:	a8 0f                	test   $0xf,%al
   1805362c4:	75 1b                	jne    0x1805362e1
   1805362c6:	66 66 0f 1f 84 00 00 	data16 nopw 0x0(%rax,%rax,1)
   1805362cd:	00 00 00 
   1805362d0:	0f 11 00             	movups %xmm0,(%rax)
   1805362d3:	48 8d 40 10          	lea    0x10(%rax),%rax
   1805362d7:	48 83 ea 01          	sub    $0x1,%rdx
   1805362db:	75 f3                	jne    0x1805362d0
   1805362dd:	49 8b c0             	mov    %r8,%rax
   1805362e0:	c3                   	ret
   1805362e1:	0f 11 00             	movups %xmm0,(%rax)
   1805362e4:	48 8d 40 10          	lea    0x10(%rax),%rax
   1805362e8:	48 83 ea 01          	sub    $0x1,%rdx
   1805362ec:	75 f3                	jne    0x1805362e1
   1805362ee:	49 8b c0             	mov    %r8,%rax
   1805362f1:	c3                   	ret
   1805362f2:	cc                   	int3
   1805362f3:	cc                   	int3
   1805362f4:	cc                   	int3
   1805362f5:	cc                   	int3
   1805362f6:	cc                   	int3
   1805362f7:	cc                   	int3
   1805362f8:	cc                   	int3
   1805362f9:	cc                   	int3
   1805362fa:	cc                   	int3
   1805362fb:	cc                   	int3
   1805362fc:	cc                   	int3
   1805362fd:	cc                   	int3
   1805362fe:	cc                   	int3
   1805362ff:	cc                   	int3
   180536300:	48 89 5c 24 08       	mov    %rbx,0x8(%rsp)
   180536305:	48 89 6c 24 18       	mov    %rbp,0x18(%rsp)
   18053630a:	48 89 74 24 20       	mov    %rsi,0x20(%rsp)
   18053630f:	48 89 54 24 10       	mov    %rdx,0x10(%rsp)
   180536314:	57                   	push   %rdi
   180536315:	41 54                	push   %r12
   180536317:	41 55                	push   %r13
   180536319:	41 56                	push   %r14
   18053631b:	41 57                	push   %r15
   18053631d:	48 83 ec 50          	sub    $0x50,%rsp
   180536321:	48 63 bc 24 a0 00 00 	movslq 0xa0(%rsp),%rdi
   180536328:	00 
   180536329:	48 8d 15 88 aa 11 02 	lea    0x211aa88(%rip),%rdx        # 0x182650db8
   180536330:	4c 8b b4 24 b0 00 00 	mov    0xb0(%rsp),%r14
   180536337:	00 
   180536338:	49 8b e9             	mov    %r9,%rbp
   18053633b:	0f 29 74 24 40       	movaps %xmm6,0x40(%rsp)
   180536340:	49 8b d8             	mov    %r8,%rbx
   180536343:	66 0f 6e b1 80 00 24 	movd   0x240080(%rcx),%xmm6
   18053634a:	00 
   18053634b:	8d 04 fd 00 00 00 00 	lea    0x0(,%rdi,8),%eax
   180536352:	4c 63 d0             	movslq %eax,%r10
   180536355:	8d 04 3f             	lea    (%rdi,%rdi,1),%eax
   180536358:	0f 5b f6             	cvtdq2ps %xmm6,%xmm6
   18053635b:	4f 8d 24 96          	lea    (%r14,%r10,4),%r12
   18053635f:	4c 63 d0             	movslq %eax,%r10
   180536362:	4d 8d 3c bc          	lea    (%r12,%rdi,4),%r15
   180536366:	4c 89 a4 24 b0 00 00 	mov    %r12,0xb0(%rsp)
   18053636d:	00 
   18053636e:	f3 0f 59 71 24       	mulss  0x24(%rcx),%xmm6
   180536373:	48 8d 0d 3e aa 11 02 	lea    0x211aa3e(%rip),%rcx        # 0x182650db8
   18053637a:	4b 8d 34 97          	lea    (%r15,%r10,4),%rsi
   18053637e:	4e 8d 2c 96          	lea    (%rsi,%r10,4),%r13
   180536382:	ff 15 80 4c 67 01    	call   *0x1674c80(%rip)        # 0x181bab008
   180536388:	f2 0f 10 0d b8 de f8 	movsd  0x1f8deb8(%rip),%xmm1        # 0x1824c4248
   18053638f:	01 
   180536390:	44 8b cf             	mov    %edi,%r9d
   180536393:	0f 5a c6             	cvtps2pd %xmm6,%xmm0
   180536396:	85 c0                	test   %eax,%eax
   180536398:	48 8b d3             	mov    %rbx,%rdx
   18053639b:	49 8b cc             	mov    %r12,%rcx
   18053639e:	0f 94 84 24 a8 00 00 	sete   0xa8(%rsp)
   1805363a5:	00 
   1805363a6:	f2 0f 5e c8          	divsd  %xmm0,%xmm1
   1805363aa:	66 0f 5a d1          	cvtpd2ps %xmm1,%xmm2
   1805363ae:	e8 4d 76 ff ff       	call   0x18052da00
   1805363b3:	80 bd 10 08 00 00 00 	cmpb   $0x0,0x810(%rbp)
   1805363ba:	44 8b c7             	mov    %edi,%r8d
   1805363bd:	74 1a                	je     0x1805363d9
   1805363bf:	f3 0f 10 0d dd da f8 	movss  0x1f8dadd(%rip),%xmm1        # 0x1824c3ea4
   1805363c6:	01 
   1805363c7:	48 8b 8c 24 88 00 00 	mov    0x88(%rsp),%rcx
   1805363ce:	00 
   1805363cf:	e8 7c 77 ff ff       	call   0x18052db50
   1805363d4:	e9 88 01 00 00       	jmp    0x180536561
   1805363d9:	f3 0f 10 35 c3 da f8 	movss  0x1f8dac3(%rip),%xmm6        # 0x1824c3ea4
   1805363e0:	01 
   1805363e1:	48 8b ce             	mov    %rsi,%rcx
   1805363e4:	0f 28 ce             	movaps %xmm6,%xmm1
   1805363e7:	e8 64 77 ff ff       	call   0x18052db50
   1805363ec:	44 8b c7             	mov    %edi,%r8d
   1805363ef:	0f 57 c9             	xorps  %xmm1,%xmm1
   1805363f2:	49 8b cd             	mov    %r13,%rcx
   1805363f5:	e8 56 77 ff ff       	call   0x18052db50
   1805363fa:	33 db                	xor    %ebx,%ebx
   1805363fc:	39                   	.byte 0x39
   1805363fd:	9d                   	popf
   1805363fe:	14 08                	adc    $0x8,%al
@@ -0,0 +1,68 @@
 /tmp/det/ctor_dc30.bin:     file format binary
 Disassembly of section .data:
 000000018052dc30 <.data>:
   18052dc30:	rex push rbx
   18052dc32:	sub    rsp,0x50
   18052dc36:	mov    rbx,rcx
   18052dc39:	cmp    DWORD PTR [rcx+0x4],edx
   18052dc3c:	je     0x18052dd1d
   18052dc42:	mov    DWORD PTR [rcx+0x8],r8d
   18052dc46:	mov    r8d,edx
   18052dc49:	mov    DWORD PTR [rcx+0x4],edx
   18052dc4c:	lea    rdx,[rsp+0x30]
   18052dc51:	mov    QWORD PTR [rsp+0x60],rsi
   18052dc56:	mov    QWORD PTR [rsp+0x68],rdi
   18052dc5b:	call   0x1805335c0
   18052dc60:	mov    r8d,DWORD PTR [rbx+0x8]
   18052dc64:	test   r8d,r8d
   18052dc67:	jns    0x18052dc8c
   18052dc69:	mov    edx,DWORD PTR [rsp+0x34]
   18052dc6d:	lea    rdi,[rbx+0x48]
   18052dc71:	mov    rcx,rdi
   18052dc74:	call   0x180534550
   18052dc79:	mov    edx,DWORD PTR [rsp+0x38]
   18052dc7d:	lea    rcx,[rbx+0x68]
   18052dc81:	call   0x180534550
   18052dc86:	mov    edx,DWORD PTR [rsp+0x3c]
   18052dc8a:	jmp    0x18052dcba
   18052dc8c:	lea    rdx,[rsp+0x40]
   18052dc91:	mov    rcx,rbx
   18052dc94:	call   0x1805335c0
   18052dc99:	mov    edx,DWORD PTR [rsp+0x44]
   18052dc9d:	lea    rdi,[rbx+0x48]
   18052dca1:	mov    rcx,rdi
   18052dca4:	call   0x180534550
   18052dca9:	mov    edx,DWORD PTR [rsp+0x48]
   18052dcad:	lea    rcx,[rbx+0x68]
   18052dcb1:	call   0x180534550
   18052dcb6:	mov    edx,DWORD PTR [rsp+0x4c]
   18052dcba:	lea    rsi,[rbx+0x58]
   18052dcbe:	mov    rcx,rsi
   18052dcc1:	call   0x180534550
   18052dcc6:	lea    rdx,[rip+0x21230eb]        # 0x182650db8
   18052dccd:	lea    rcx,[rip+0x21230e4]        # 0x182650db8
   18052dcd4:	call   QWORD PTR [rip+0x167d32e]        # 0x181bab008
   18052dcda:	mov    r8d,DWORD PTR [rbx]
   18052dcdd:	mov    r9d,0x1
   18052dce3:	mov    edx,DWORD PTR [rsp+0x30]
   18052dce7:	test   eax,eax
   18052dce9:	mov    rax,QWORD PTR [rsi]
   18052dcec:	mov    QWORD PTR [rsp+0x28],rax
   18052dcf1:	mov    rax,QWORD PTR [rdi]
   18052dcf4:	mov    QWORD PTR [rsp+0x20],rax
   18052dcf9:	jne    0x18052dd06
   18052dcfb:	lea    rcx,[rbx+0x18]
   18052dcff:	call   0x180002240
   18052dd04:	jmp    0x18052dd0f
   18052dd06:	lea    rcx,[rbx+0x20]
   18052dd0a:	call   0x180001c10
   18052dd0f:	mov    rdi,QWORD PTR [rsp+0x68]
   18052dd14:	mov    rsi,QWORD PTR [rsp+0x60]
   18052dd19:	mov    BYTE PTR [rbx+0x15],0x1
   18052dd1d:	add    rsp,0x50
   18052dd21:	pop    rbx
   18052dd22:	ret
@@ -0,0 +1,582 @@
 /tmp/det/det_536f90.bin:     file format binary
 Disassembly of section .data:
 0000000180536f90 <.data>:
   180536f90:	4c 8b dc             	mov    %rsp,%r11
   180536f93:	55                   	push   %rbp
   180536f94:	56                   	push   %rsi
   180536f95:	57                   	push   %rdi
   180536f96:	41 54                	push   %r12
   180536f98:	41 55                	push   %r13
   180536f9a:	41 56                	push   %r14
   180536f9c:	41 57                	push   %r15
   180536f9e:	48 81 ec c0 00 00 00 	sub    $0xc0,%rsp
   180536fa5:	41 0f 29 73 b8       	movaps %xmm6,-0x48(%r11)
   180536faa:	41 0f 29 7b a8       	movaps %xmm7,-0x58(%r11)
   180536faf:	45 0f 29 43 98       	movaps %xmm8,-0x68(%r11)
   180536fb4:	45 0f 29 4b 88       	movaps %xmm9,-0x78(%r11)
   180536fb9:	48 8b 05 b0 e9 0d 02 	mov    0x20de9b0(%rip),%rax        # 0x182615970
   180536fc0:	48 33 c4             	xor    %rsp,%rax
   180536fc3:	48 89 44 24 70       	mov    %rax,0x70(%rsp)
   180536fc8:	48 63 b4 24 20 01 00 	movslq 0x120(%rsp),%rsi
   180536fcf:	00 
   180536fd0:	4c 8b f1             	mov    %rcx,%r14
   180536fd3:	49 8b 00             	mov    (%r8),%rax
   180536fd6:	4c 8b e6             	mov    %rsi,%r12
   180536fd9:	49 8b 48 08          	mov    0x8(%r8),%rcx
   180536fdd:	4c 8b fa             	mov    %rdx,%r15
   180536fe0:	f2 44 0f 10 0d 57 d1 	movsd  0x1f8d157(%rip),%xmm9        # 0x1824c4140
   180536fe7:	f8 01 
   180536fe9:	0f 57 f6             	xorps  %xmm6,%xmm6
   180536fec:	49 c1 e4 04          	shl    $0x4,%r12
   180536ff0:	44 8d 2c 36          	lea    (%rsi,%rsi,1),%r13d
   180536ff4:	0f 10 40 08          	movups 0x8(%rax),%xmm0
   180536ff8:	49 63 ed             	movslq %r13d,%rbp
   180536ffb:	4c 03 e2             	add    %rdx,%r12
   180536ffe:	0f 10 49 08          	movups 0x8(%rcx),%xmm1
   180537002:	48 c1 e5 04          	shl    $0x4,%rbp
   180537006:	f2 0f 10 38          	movsd  (%rax),%xmm7
   18053700a:	48 03 ea             	add    %rdx,%rbp
   18053700d:	f2 44 0f 10 01       	movsd  (%rcx),%xmm8
   180537012:	33 ff                	xor    %edi,%edi
   180537014:	f2 0f 11 7c 24 40    	movsd  %xmm7,0x40(%rsp)
   18053701a:	f2 44 0f 11 44 24 58 	movsd  %xmm8,0x58(%rsp)
   180537021:	4c 89 4c 24 20       	mov    %r9,0x20(%rsp)
   180537026:	0f 11 44 24 48       	movups %xmm0,0x48(%rsp)
   18053702b:	49 89 5b 18          	mov    %rbx,0x18(%r11)
   18053702f:	0f 11 4c 24 60       	movups %xmm1,0x60(%rsp)
   180537034:	48 85 ff             	test   %rdi,%rdi
   180537037:	75 40                	jne    0x180537079
   180537039:	0f 28 c7             	movaps %xmm7,%xmm0
   18053703c:	48 8d 4c 24 30       	lea    0x30(%rsp),%rcx
   180537041:	66 0f 14 c6          	unpcklpd %xmm6,%xmm0
   180537045:	44 8b c6             	mov    %esi,%r8d
   180537048:	49 8b d7             	mov    %r15,%rdx
   18053704b:	66 0f 7f 44 24 30    	movdqa %xmm0,0x30(%rsp)
   180537051:	e8 da a9 ac ff       	call   0x180001a30
   180537056:	41 0f 28 c0          	movaps %xmm8,%xmm0
   18053705a:	48 8d 4c 24 30       	lea    0x30(%rsp),%rcx
   18053705f:	66 0f 14 c6          	unpcklpd %xmm6,%xmm0
   180537063:	44 8b c6             	mov    %esi,%r8d
   180537066:	49 8b d4             	mov    %r12,%rdx
   180537069:	66 0f 7f 44 24 30    	movdqa %xmm0,0x30(%rsp)
   18053706f:	e8 bc a9 ac ff       	call   0x180001a30
   180537074:	e9 88 00 00 00       	jmp    0x180537101
   180537079:	48 83 ff 01          	cmp    $0x1,%rdi
   18053707d:	75 49                	jne    0x1805370c8
   18053707f:	8d 04 76             	lea    (%rsi,%rsi,2),%eax
   180537082:	44 8b c6             	mov    %esi,%r8d
   180537085:	48 63 d8             	movslq %eax,%rbx
   180537088:	41 0f 28 c1          	movaps %xmm9,%xmm0
   18053708c:	48 c1 e3 04          	shl    $0x4,%rbx
   180537090:	49 03 df             	add    %r15,%rbx
   180537093:	48 8b d3             	mov    %rbx,%rdx
   180537096:	e8 35 b2 ac ff       	call   0x1800022d0
   18053709b:	48 8b 54 24 20       	mov    0x20(%rsp),%rdx
   1805370a0:	44 8b ce             	mov    %esi,%r9d
   1805370a3:	4d 8b c6             	mov    %r14,%r8
   1805370a6:	48 8b cb             	mov    %rbx,%rcx
   1805370a9:	e8 d2 ad ac ff       	call   0x180001e80
   1805370ae:	8b d6                	mov    %esi,%edx
   1805370b0:	49 8b ce             	mov    %r14,%rcx
   1805370b3:	e8 58 a8 ac ff       	call   0x180001910
   1805370b8:	44 8b c6             	mov    %esi,%r8d
   1805370bb:	48 8b d5             	mov    %rbp,%rdx
   1805370be:	49 8b ce             	mov    %r14,%rcx
   1805370c1:	e8 fa a9 ac ff       	call   0x180001ac0
   1805370c6:	eb 11                	jmp    0x1805370d9
   1805370c8:	44 8b ce             	mov    %esi,%r9d
   1805370cb:	4c 8b c5             	mov    %rbp,%r8
   1805370ce:	48 8b d5             	mov    %rbp,%rdx
   1805370d1:	49 8b ce             	mov    %r14,%rcx
   1805370d4:	e8 77 ad ac ff       	call   0x180001e50
   1805370d9:	f2 0f 10 4c fc 40    	movsd  0x40(%rsp,%rdi,8),%xmm1
   1805370df:	45 8b cd             	mov    %r13d,%r9d
   1805370e2:	4d 8b c7             	mov    %r15,%r8
   1805370e5:	48 8b cd             	mov    %rbp,%rcx
   1805370e8:	e8 63 b0 ac ff       	call   0x180002150
   1805370ed:	f2 0f 10 4c fc 58    	movsd  0x58(%rsp,%rdi,8),%xmm1
   1805370f3:	45 8b cd             	mov    %r13d,%r9d
   1805370f6:	4d 8b c4             	mov    %r12,%r8
   1805370f9:	48 8b cd             	mov    %rbp,%rcx
   1805370fc:	e8 4f b0 ac ff       	call   0x180002150
   180537101:	48 ff c7             	inc    %rdi
   180537104:	48 83 ff 03          	cmp    $0x3,%rdi
   180537108:	0f 8c 26 ff ff ff    	jl     0x180537034
   18053710e:	44 8b ce             	mov    %esi,%r9d
   180537111:	4d 8b c6             	mov    %r14,%r8
   180537114:	49 8b d7             	mov    %r15,%rdx
   180537117:	49 8b cc             	mov    %r12,%rcx
   18053711a:	e8 21 ae ac ff       	call   0x180001f40
   18053711f:	48 8b 9c 24 10 01 00 	mov    0x110(%rsp),%rbx
   180537126:	00 
   180537127:	48 8b 4c 24 70       	mov    0x70(%rsp),%rcx
   18053712c:	48 33 cc             	xor    %rsp,%rcx
   18053712f:	e8 9c 9f bf 00       	call   0x1811310d0
   180537134:	4c 8d 9c 24 c0 00 00 	lea    0xc0(%rsp),%r11
   18053713b:	00 
   18053713c:	41 0f 28 73 f0       	movaps -0x10(%r11),%xmm6
   180537141:	41 0f 28 7b e0       	movaps -0x20(%r11),%xmm7
   180537146:	45 0f 28 43 d0       	movaps -0x30(%r11),%xmm8
   18053714b:	45 0f 28 4b c0       	movaps -0x40(%r11),%xmm9
   180537150:	49 8b e3             	mov    %r11,%rsp
   180537153:	41 5f                	pop    %r15
   180537155:	41 5e                	pop    %r14
   180537157:	41 5d                	pop    %r13
   180537159:	41 5c                	pop    %r12
   18053715b:	5f                   	pop    %rdi
   18053715c:	5e                   	pop    %rsi
   18053715d:	5d                   	pop    %rbp
   18053715e:	c3                   	ret
   18053715f:	cc                   	int3
   180537160:	48 89 5c 24 08       	mov    %rbx,0x8(%rsp)
   180537165:	48 89 54 24 10       	mov    %rdx,0x10(%rsp)
   18053716a:	57                   	push   %rdi
   18053716b:	48 83 ec 40          	sub    $0x40,%rsp
   18053716f:	48 8b fa             	mov    %rdx,%rdi
   180537172:	0f 29 74 24 30       	movaps %xmm6,0x30(%rsp)
   180537177:	f3 41 0f 10 30       	movss  (%r8),%xmm6
   18053717c:	41 b9 0b 00 00 00    	mov    $0xb,%r9d
   180537182:	4c 8d 05 a7 48 f7 01 	lea    0x1f748a7(%rip),%r8        # 0x1824aba30
   180537189:	ba dd 7d 54 ff       	mov    $0xff547ddd,%edx
   18053718e:	48 8b d9             	mov    %rcx,%rbx
   180537191:	e8 6a 76 c0 00       	call   0x18113e800
   180537196:	48 8b 4b 08          	mov    0x8(%rbx),%rcx
   18053719a:	48 8b d0             	mov    %rax,%rdx
   18053719d:	e8 be 38 c6 00       	call   0x18119aa60
   1805371a2:	8b 80 88 01 00 00    	mov    0x188(%rax),%eax
   1805371a8:	89 44 24 58          	mov    %eax,0x58(%rsp)
   1805371ac:	0f 57 d2             	xorps  %xmm2,%xmm2
   1805371af:	f3 0f 10 44 24 58    	movss  0x58(%rsp),%xmm0
   1805371b5:	0f 28 ce             	movaps %xmm6,%xmm1
   1805371b8:	0f 2f c2             	comiss %xmm2,%xmm0
   1805371bb:	48 8b cf             	mov    %rdi,%rcx
   1805371be:	41 0f 97 c0          	seta   %r8b
   1805371c2:	e8 e9 36 ee 00       	call   0x18141a8b0
   1805371c7:	48 8b 5c 24 50       	mov    0x50(%rsp),%rbx
   1805371cc:	48 8b c7             	mov    %rdi,%rax
   1805371cf:	0f 28 74 24 30       	movaps 0x30(%rsp),%xmm6
   1805371d4:	48 83 c4 40          	add    $0x40,%rsp
   1805371d8:	5f                   	pop    %rdi
   1805371d9:	c3                   	ret
   1805371da:	cc                   	int3
   1805371db:	cc                   	int3
   1805371dc:	cc                   	int3
   1805371dd:	cc                   	int3
   1805371de:	cc                   	int3
   1805371df:	cc                   	int3
   1805371e0:	4c 8b dc             	mov    %rsp,%r11
   1805371e3:	53                   	push   %rbx
   1805371e4:	48 83 ec 50          	sub    $0x50,%rsp
   1805371e8:	49 c7 43 d8 fe ff ff 	movq   $0xfffffffffffffffe,-0x28(%r11)
   1805371ef:	ff 
   1805371f0:	0f 29 74 24 40       	movaps %xmm6,0x40(%rsp)
   1805371f5:	48 8b d9             	mov    %rcx,%rbx
   1805371f8:	48 8b 02             	mov    (%rdx),%rax
   1805371fb:	4c 8d 05 96 4d ca 01 	lea    0x1ca4d96(%rip),%r8        # 0x1821dbf98
   180537202:	4c 89 02             	mov    %r8,(%rdx)
   180537205:	49 8d 4b 08          	lea    0x8(%r11),%rcx
   180537209:	49 89 4b 18          	mov    %rcx,0x18(%r11)
   18053720d:	49 8d 4b 10          	lea    0x10(%r11),%rcx
   180537211:	49 89 4b 20          	mov    %rcx,0x20(%r11)
   180537215:	49 89 43 10          	mov    %rax,0x10(%r11)
   180537219:	4d 89 43 08          	mov    %r8,0x8(%r11)
   18053721d:	41 b9 0b 00 00 00    	mov    $0xb,%r9d
   180537223:	4c 8d 05 06 48 f7 01 	lea    0x1f74806(%rip),%r8        # 0x1824aba30
   18053722a:	ba dd 7d 54 ff       	mov    $0xff547ddd,%edx
   18053722f:	e8 cc 75 c0 00       	call   0x18113e800
   180537234:	48 89 44 24 78       	mov    %rax,0x78(%rsp)
   180537239:	48 8b d0             	mov    %rax,%rdx
   18053723c:	48 8b 4b 08          	mov    0x8(%rbx),%rcx
   180537240:	e8 1b 38 c6 00       	call   0x18119aa60
   180537245:	8b 80 88 01 00 00    	mov    0x188(%rax),%eax
   18053724b:	48 8d 4c 24 68       	lea    0x68(%rsp),%rcx
   180537250:	e8 0b 34 ee 00       	call   0x18141a660
   180537255:	0f 28 f0             	movaps %xmm0,%xmm6
   180537258:	48 8b 4c 24 60       	mov    0x60(%rsp),%rcx
   18053725d:	48 83 c1 f0          	add    $0xfffffffffffffff0,%rcx
   180537261:	8b 01                	mov    (%rcx),%eax
   180537263:	a9 00 00 00 30       	test   $0x30000000,%eax
   180537268:	75 18                	jne    0x180537282
   18053726a:	b8 ff ff ff ff       	mov    $0xffffffff,%eax
   18053726f:	f0 0f c1 01          	lock xadd %eax,(%rcx)
   180537273:	ff c8                	dec    %eax
   180537275:	83 f8 ff             	cmp    $0xffffffff,%eax
   180537278:	75 08                	jne    0x180537282
   18053727a:	e8 75 9e bf 00       	call   0x1811310f4
   18053727f:	0f 28 c6             	movaps %xmm6,%xmm0
   180537282:	0f 28 74 24 40       	movaps 0x40(%rsp),%xmm6
   180537287:	48 83 c4 50          	add    $0x50,%rsp
   18053728b:	5b                   	pop    %rbx
   18053728c:	c3                   	ret
   18053728d:	cc                   	int3
   18053728e:	cc                   	int3
   18053728f:	cc                   	int3
   180537290:	40 53                	rex push %rbx
   180537292:	48 83 ec 20          	sub    $0x20,%rsp
   180537296:	f3 41 0f 10 08       	movss  (%r8),%xmm1
   18053729b:	45 33 c9             	xor    %r9d,%r9d
   18053729e:	0f 5a c9             	cvtps2pd %xmm1,%xmm1
   1805372a1:	48 8b ca             	mov    %rdx,%rcx
   1805372a4:	48 8b da             	mov    %rdx,%rbx
   1805372a7:	45 8d 41 01          	lea    0x1(%r9),%r8d
   1805372ab:	e8 f0 c6 cf 00       	call   0x1812339a0
   1805372b0:	48 8b c3             	mov    %rbx,%rax
   1805372b3:	48 83 c4 20          	add    $0x20,%rsp
   1805372b7:	5b                   	pop    %rbx
   1805372b8:	c3                   	ret
   1805372b9:	cc                   	int3
   1805372ba:	cc                   	int3
   1805372bb:	cc                   	int3
   1805372bc:	cc                   	int3
   1805372bd:	cc                   	int3
   1805372be:	cc                   	int3
   1805372bf:	cc                   	int3
   1805372c0:	40 53                	rex push %rbx
   1805372c2:	48 83 ec 30          	sub    $0x30,%rsp
   1805372c6:	f3 41 0f 10 10       	movss  (%r8),%xmm2
   1805372cb:	48 8b da             	mov    %rdx,%rbx
   1805372ce:	f2 0f 10 05 92 cc f8 	movsd  0x1f8cc92(%rip),%xmm0        # 0x1824c3f68
   1805372d5:	01 
   1805372d6:	0f 5a ca             	cvtps2pd %xmm2,%xmm1
   1805372d9:	66 0f 2f c1          	comisd %xmm1,%xmm0
   1805372dd:	76 2b                	jbe    0x18053730a
   1805372df:	4c 8d 05 5e 3b f7 01 	lea    0x1f73b5e(%rip),%r8        # 0x1824aae44
   1805372e6:	ba cc 37 ce 4a       	mov    $0x4ace37cc,%edx
   1805372eb:	41 b9 04 00 00 00    	mov    $0x4,%r9d
   1805372f1:	e8 0a 75 c0 00       	call   0x18113e800
   1805372f6:	48 8b d0             	mov    %rax,%rdx
   1805372f9:	48 8b cb             	mov    %rbx,%rcx
   1805372fc:	e8 df 51 f6 ff       	call   0x18049c4e0
   180537301:	48 8b c3             	mov    %rbx,%rax
   180537304:	48 83 c4 30          	add    $0x30,%rsp
   180537308:	5b                   	pop    %rbx
   180537309:	c3                   	ret
   18053730a:	0f 5a ca             	cvtps2pd %xmm2,%xmm1
   18053730d:	66 0f 2f 0d 53 cf f8 	comisd 0x1f8cf53(%rip),%xmm1        # 0x1824c4268
   180537314:	01 
   180537315:	76 0e                	jbe    0x180537325
   180537317:	4c 8d 05 f6 42 f7 01 	lea    0x1f742f6(%rip),%r8        # 0x1824ab614
   18053731e:	ba d3 10 24 91       	mov    $0x912410d3,%edx
   180537323:	eb c6                	jmp    0x1805372eb
   180537325:	45 33 c9             	xor    %r9d,%r9d
   180537328:	48 8b cb             	mov    %rbx,%rcx
   18053732b:	45 8d 41 01          	lea    0x1(%r9),%r8d
   18053732f:	e8 6c c6 cf 00       	call   0x1812339a0
   180537334:	48 8b c3             	mov    %rbx,%rax
   180537337:	48 83 c4 30          	add    $0x30,%rsp
   18053733b:	5b                   	pop    %rbx
   18053733c:	c3                   	ret
   18053733d:	cc                   	int3
   18053733e:	cc                   	int3
   18053733f:	cc                   	int3
   180537340:	40 53                	rex push %rbx
   180537342:	48 83 ec 30          	sub    $0x30,%rsp
   180537346:	f3 41 0f 10 10       	movss  (%r8),%xmm2
   18053734b:	48 8b da             	mov    %rdx,%rbx
   18053734e:	f2 0f 10 05 12 cc f8 	movsd  0x1f8cc12(%rip),%xmm0        # 0x1824c3f68
   180537355:	01 
   180537356:	0f 5a ca             	cvtps2pd %xmm2,%xmm1
   180537359:	66 0f 2f c1          	comisd %xmm1,%xmm0
   18053735d:	76 2b                	jbe    0x18053738a
   18053735f:	4c 8d 05 06 4d f7 01 	lea    0x1f74d06(%rip),%r8        # 0x1824ac06c
   180537366:	ba b2 b5 fd e6       	mov    $0xe6fdb5b2,%edx
   18053736b:	41 b9 04 00 00 00    	mov    $0x4,%r9d
   180537371:	e8 8a 74 c0 00       	call   0x18113e800
   180537376:	48 8b d0             	mov    %rax,%rdx
   180537379:	48 8b cb             	mov    %rbx,%rcx
   18053737c:	e8 5f 51 f6 ff       	call   0x18049c4e0
   180537381:	48 8b c3             	mov    %rbx,%rax
   180537384:	48 83 c4 30          	add    $0x30,%rsp
   180537388:	5b                   	pop    %rbx
   180537389:	c3                   	ret
   18053738a:	0f 5a ca             	cvtps2pd %xmm2,%xmm1
   18053738d:	66 0f 2f 0d d3 ce f8 	comisd 0x1f8ced3(%rip),%xmm1        # 0x1824c4268
   180537394:	01 
   180537395:	76 0e                	jbe    0x1805373a5
   180537397:	4c 8d 05 3a 51 f7 01 	lea    0x1f7513a(%rip),%r8        # 0x1824ac4d8
   18053739e:	ba 56 9f 92 12       	mov    $0x12929f56,%edx
   1805373a3:	eb c6                	jmp    0x18053736b
   1805373a5:	45 33 c9             	xor    %r9d,%r9d
   1805373a8:	48 8b cb             	mov    %rbx,%rcx
   1805373ab:	45 8d 41 01          	lea    0x1(%r9),%r8d
   1805373af:	e8 ec c5 cf 00       	call   0x1812339a0
   1805373b4:	48 8b c3             	mov    %rbx,%rax
   1805373b7:	48 83 c4 30          	add    $0x30,%rsp
   1805373bb:	5b                   	pop    %rbx
   1805373bc:	c3                   	ret
   1805373bd:	cc                   	int3
   1805373be:	cc                   	int3
   1805373bf:	cc                   	int3
   1805373c0:	f3 0f 10 0a          	movss  (%rdx),%xmm1
   1805373c4:	0f 57 c0             	xorps  %xmm0,%xmm0
   1805373c7:	0f 2e c8             	ucomiss %xmm0,%xmm1
   1805373ca:	7a 06                	jp     0x1805373d2
   1805373cc:	75 04                	jne    0x1805373d2
   1805373ce:	b2 01                	mov    $0x1,%dl
   1805373d0:	eb 02                	jmp    0x1805373d4
   1805373d2:	33 d2                	xor    %edx,%edx
   1805373d4:	48 8b 41 08          	mov    0x8(%rcx),%rax
   1805373d8:	48 8b 88 68 01 00 00 	mov    0x168(%rax),%rcx
   1805373df:	48 8b 01             	mov    (%rcx),%rax
   1805373e2:	48 ff a0 b8 00 00 00 	rex.W jmp *0xb8(%rax)
   1805373e9:	cc                   	int3
   1805373ea:	cc                   	int3
   1805373eb:	cc                   	int3
   1805373ec:	cc                   	int3
   1805373ed:	cc                   	int3
   1805373ee:	cc                   	int3
   1805373ef:	cc                   	int3
   1805373f0:	48 8b 41 08          	mov    0x8(%rcx),%rax
   1805373f4:	f3 0f 10 0a          	movss  (%rdx),%xmm1
   1805373f8:	48 8b 88 68 01 00 00 	mov    0x168(%rax),%rcx
   1805373ff:	48 8b 01             	mov    (%rcx),%rax
   180537402:	48 ff 60 60          	rex.W jmp *0x60(%rax)
   180537406:	cc                   	int3
   180537407:	cc                   	int3
   180537408:	cc                   	int3
   180537409:	cc                   	int3
   18053740a:	cc                   	int3
   18053740b:	cc                   	int3
   18053740c:	cc                   	int3
   18053740d:	cc                   	int3
   18053740e:	cc                   	int3
   18053740f:	cc                   	int3
   180537410:	48 8b 41 08          	mov    0x8(%rcx),%rax
   180537414:	f3 0f 10 0a          	movss  (%rdx),%xmm1
   180537418:	f3 0f 59 0d 98 c8 f8 	mulss  0x1f8c898(%rip),%xmm1        # 0x1824c3cb8
   18053741f:	01 
   180537420:	48 8b 88 68 01 00 00 	mov    0x168(%rax),%rcx
   180537427:	48 8b 01             	mov    (%rcx),%rax
   18053742a:	48 ff 60 68          	rex.W jmp *0x68(%rax)
   18053742e:	cc                   	int3
   18053742f:	cc                   	int3
   180537430:	48 8b 41 08          	mov    0x8(%rcx),%rax
   180537434:	f3 0f 10 0a          	movss  (%rdx),%xmm1
   180537438:	f3 0f 59 0d 78 c8 f8 	mulss  0x1f8c878(%rip),%xmm1        # 0x1824c3cb8
   18053743f:	01 
   180537440:	48 8b 88 68 01 00 00 	mov    0x168(%rax),%rcx
   180537447:	48 8b 01             	mov    (%rcx),%rax
   18053744a:	48 ff 60 70          	rex.W jmp *0x70(%rax)
   18053744e:	cc                   	int3
   18053744f:	cc                   	int3
   180537450:	48 89 5c 24 08       	mov    %rbx,0x8(%rsp)
   180537455:	48 89 74 24 10       	mov    %rsi,0x10(%rsp)
   18053745a:	57                   	push   %rdi
   18053745b:	48 83 ec 30          	sub    $0x30,%rsp
   18053745f:	f3 0f 10 05 dd cf f8 	movss  0x1f8cfdd(%rip),%xmm0        # 0x1824c4444
   180537466:	01 
   180537467:	48 8b fa             	mov    %rdx,%rdi
   18053746a:	0f 29 74 24 20       	movaps %xmm6,0x20(%rsp)
   18053746f:	48 8b d9             	mov    %rcx,%rbx
   180537472:	e8 59 d8 4d 01       	call   0x181a14cd0
   180537477:	48 8b 43 08          	mov    0x8(%rbx),%rax
   18053747b:	0f 28 f0             	movaps %xmm0,%xmm6
   18053747e:	f3 0f 10 05 6a d1 f8 	movss  0x1f8d16a(%rip),%xmm0        # 0x1824c45f0
   180537485:	01 
   180537486:	48 8b 98 68 01 00 00 	mov    0x168(%rax),%rbx
   18053748d:	48 8b 33             	mov    (%rbx),%rsi
   180537490:	e8 3b d8 4d 01       	call   0x181a14cd0
   180537495:	f3 0f 10 0f          	movss  (%rdi),%xmm1
   180537499:	f3 0f 5c c6          	subss  %xmm6,%xmm0
   18053749d:	f3 0f 59 0d 13 c8 f8 	mulss  0x1f8c813(%rip),%xmm1        # 0x1824c3cb8
   1805374a4:	01 
   1805374a5:	f3 0f 59 c1          	mulss  %xmm1,%xmm0
   1805374a9:	f3 0f 58 c6          	addss  %xmm6,%xmm0
   1805374ad:	e8 fa d7 4d 01       	call   0x181a14cac
   1805374b2:	0f 28 c8             	movaps %xmm0,%xmm1
   1805374b5:	48 8b cb             	mov    %rbx,%rcx
   1805374b8:	48 8b 46 78          	mov    0x78(%rsi),%rax
   1805374bc:	48 8b 5c 24 40       	mov    0x40(%rsp),%rbx
   1805374c1:	48 8b 74 24 48       	mov    0x48(%rsp),%rsi
   1805374c6:	0f 28 74 24 20       	movaps 0x20(%rsp),%xmm6
   1805374cb:	48 83 c4 30          	add    $0x30,%rsp
   1805374cf:	5f                   	pop    %rdi
   1805374d0:	48 ff e0             	rex.W jmp *%rax
   1805374d3:	cc                   	int3
   1805374d4:	cc                   	int3
   1805374d5:	cc                   	int3
   1805374d6:	cc                   	int3
   1805374d7:	cc                   	int3
   1805374d8:	cc                   	int3
   1805374d9:	cc                   	int3
   1805374da:	cc                   	int3
   1805374db:	cc                   	int3
   1805374dc:	cc                   	int3
   1805374dd:	cc                   	int3
   1805374de:	cc                   	int3
   1805374df:	cc                   	int3
   1805374e0:	48 89 5c 24 08       	mov    %rbx,0x8(%rsp)
   1805374e5:	48 89 74 24 10       	mov    %rsi,0x10(%rsp)
   1805374ea:	57                   	push   %rdi
   1805374eb:	48 83 ec 40          	sub    $0x40,%rsp
   1805374ef:	f3 0f 10 05 f1 ce f8 	movss  0x1f8cef1(%rip),%xmm0        # 0x1824c43e8
   1805374f6:	01 
   1805374f7:	48 8b fa             	mov    %rdx,%rdi
   1805374fa:	0f 29 74 24 30       	movaps %xmm6,0x30(%rsp)
   1805374ff:	48 8b d9             	mov    %rcx,%rbx
   180537502:	0f 29 7c 24 20       	movaps %xmm7,0x20(%rsp)
   180537507:	e8 c4 d7 4d 01       	call   0x181a14cd0
   18053750c:	48 8b 43 08          	mov    0x8(%rbx),%rax
   180537510:	0f 28 f8             	movaps %xmm0,%xmm7
   180537513:	f3 0f 10 07          	movss  (%rdi),%xmm0
   180537517:	f3 0f 59 05 99 c7 f8 	mulss  0x1f8c799(%rip),%xmm0        # 0x1824c3cb8
   18053751e:	01 
   18053751f:	48 8b 98 68 01 00 00 	mov    0x168(%rax),%rbx
   180537526:	f3 0f 10 0d ca c9 f8 	movss  0x1f8c9ca(%rip),%xmm1        # 0x1824c3ef8
   18053752d:	01 
   18053752e:	48 8b 33             	mov    (%rbx),%rsi
   180537531:	e8 a6 d7 4d 01       	call   0x181a14cdc
   180537536:	0f 28 f0             	movaps %xmm0,%xmm6
   180537539:	f3 0f 10 05 a3 d0 f8 	movss  0x1f8d0a3(%rip),%xmm0        # 0x1824c45e4
   180537540:	01 
   180537541:	e8 8a d7 4d 01       	call   0x181a14cd0
   180537546:	f3 0f 5c c7          	subss  %xmm7,%xmm0
   18053754a:	f3 0f 59 f0          	mulss  %xmm0,%xmm6
   18053754e:	f3 0f 58 f7          	addss  %xmm7,%xmm6
   180537552:	0f 28 c6             	movaps %xmm6,%xmm0
   180537555:	e8 52 d7 4d 01       	call   0x181a14cac
   18053755a:	0f 28 c8             	movaps %xmm0,%xmm1
   18053755d:	48 8b cb             	mov    %rbx,%rcx
   180537560:	48 8b 86 80 00 00 00 	mov    0x80(%rsi),%rax
   180537567:	48 8b 5c 24 50       	mov    0x50(%rsp),%rbx
   18053756c:	48 8b 74 24 58       	mov    0x58(%rsp),%rsi
   180537571:	0f 28 74 24 30       	movaps 0x30(%rsp),%xmm6
   180537576:	0f 28 7c 24 20       	movaps 0x20(%rsp),%xmm7
   18053757b:	48 83 c4 40          	add    $0x40,%rsp
   18053757f:	5f                   	pop    %rdi
   180537580:	48 ff e0             	rex.W jmp *%rax
   180537583:	cc                   	int3
   180537584:	cc                   	int3
   180537585:	cc                   	int3
   180537586:	cc                   	int3
   180537587:	cc                   	int3
   180537588:	cc                   	int3
   180537589:	cc                   	int3
   18053758a:	cc                   	int3
   18053758b:	cc                   	int3
   18053758c:	cc                   	int3
   18053758d:	cc                   	int3
   18053758e:	cc                   	int3
   18053758f:	cc                   	int3
   180537590:	48 8b 41 08          	mov    0x8(%rcx),%rax
   180537594:	48 8b 88 68 01 00 00 	mov    0x168(%rax),%rcx
   18053759b:	48 81 c1 d8 03 00 00 	add    $0x3d8,%rcx
   1805375a2:	83 79 6c 00          	cmpl   $0x0,0x6c(%rcx)
   1805375a6:	76 09                	jbe    0x1805375b1
   1805375a8:	48 8b 41 60          	mov    0x60(%rcx),%rax
   1805375ac:	4c 8b 00             	mov    (%rax),%r8
   1805375af:	eb 03                	jmp    0x1805375b4
   1805375b1:	45 33 c0             	xor    %r8d,%r8d
   1805375b4:	f3 0f 10 05 e8 c8 f8 	movss  0x1f8c8e8(%rip),%xmm0        # 0x1824c3ea4
   1805375bb:	01 
   1805375bc:	0f 2f 02             	comiss (%rdx),%xmm0
   1805375bf:	0f 97 c0             	seta   %al
   1805375c2:	41 88 80 10 08 00 00 	mov    %al,0x810(%r8)
   1805375c9:	c6 81 8f 00 24 00 01 	movb   $0x1,0x24008f(%rcx)
   1805375d0:	e9 5b 86 ff ff       	jmp    0x18052fc30
   1805375d5:	cc                   	int3
   1805375d6:	cc                   	int3
   1805375d7:	cc                   	int3
   1805375d8:	cc                   	int3
   1805375d9:	cc                   	int3
   1805375da:	cc                   	int3
   1805375db:	cc                   	int3
   1805375dc:	cc                   	int3
   1805375dd:	cc                   	int3
   1805375de:	cc                   	int3
   1805375df:	cc                   	int3
   1805375e0:	48 8b 41 08          	mov    0x8(%rcx),%rax
   1805375e4:	f3 0f 10 02          	movss  (%rdx),%xmm0
   1805375e8:	48 8b 08             	mov    (%rax),%rcx
   1805375eb:	48 81 c1 d8 03 00 00 	add    $0x3d8,%rcx
   1805375f2:	83 79 6c 00          	cmpl   $0x0,0x6c(%rcx)
   1805375f6:	76 1b                	jbe    0x180537613
   1805375f8:	48 8b 41 60          	mov    0x60(%rcx),%rax
   1805375fc:	48 8b 10             	mov    (%rax),%rdx
   1805375ff:	f3 0f 11 82 0c 08 00 	movss  %xmm0,0x80c(%rdx)
   180537606:	00 
   180537607:	c6 81 8f 00 24 00 01 	movb   $0x1,0x24008f(%rcx)
   18053760e:	e9 1d 86 ff ff       	jmp    0x18052fc30
   180537613:	33 d2                	xor    %edx,%edx
   180537615:	f3 0f 11 82 0c 08 00 	movss  %xmm0,0x80c(%rdx)
   18053761c:	00 
   18053761d:	c6 81 8f 00 24 00 01 	movb   $0x1,0x24008f(%rcx)
   180537624:	e9 07 86 ff ff       	jmp    0x18052fc30
   180537629:	cc                   	int3
   18053762a:	cc                   	int3
   18053762b:	cc                   	int3
   18053762c:	cc                   	int3
   18053762d:	cc                   	int3
   18053762e:	cc                   	int3
   18053762f:	cc                   	int3
   180537630:	48 8b 41 08          	mov    0x8(%rcx),%rax
   180537634:	f3 0f 10 02          	movss  (%rdx),%xmm0
   180537638:	48 8b 08             	mov    (%rax),%rcx
   18053763b:	48 81 c1 d8 03 00 00 	add    $0x3d8,%rcx
   180537642:	83 79 6c 00          	cmpl   $0x0,0x6c(%rcx)
   180537646:	76 1b                	jbe    0x180537663
   180537648:	48 8b 41 60          	mov    0x60(%rcx),%rax
   18053764c:	48 8b 10             	mov    (%rax),%rdx
   18053764f:	f3 0f 11 82 04 08 00 	movss  %xmm0,0x804(%rdx)
   180537656:	00 
   180537657:	c6 81 8f 00 24 00 01 	movb   $0x1,0x24008f(%rcx)
   18053765e:	e9 cd 85 ff ff       	jmp    0x18052fc30
   180537663:	33 d2                	xor    %edx,%edx
   180537665:	f3 0f 11 82 04 08 00 	movss  %xmm0,0x804(%rdx)
   18053766c:	00 
   18053766d:	c6 81 8f 00 24 00 01 	movb   $0x1,0x24008f(%rcx)
   180537674:	e9 b7 85 ff ff       	jmp    0x18052fc30
   180537679:	cc                   	int3
   18053767a:	cc                   	int3
   18053767b:	cc                   	int3
   18053767c:	cc                   	int3
   18053767d:	cc                   	int3
   18053767e:	cc                   	int3
   18053767f:	cc                   	int3
   180537680:	40 53                	rex push %rbx
   180537682:	48 83 ec 20          	sub    $0x20,%rsp
   180537686:	f3 41 0f 10 08       	movss  (%r8),%xmm1
   18053768b:	48 8b ca             	mov    %rdx,%rcx
   18053768e:	48 8b da             	mov    %rdx,%rbx
   180537691:	e8 3a 39 ee 00       	call   0x18141afd0
   180537696:	48 8b c3             	mov    %rbx,%rax
   180537699:	48 83 c4 20          	add    $0x20,%rsp
   18053769d:	5b                   	pop    %rbx
   18053769e:	c3                   	ret
   18053769f:	cc                   	int3
   1805376a0:	48 89 4c 24 08       	mov    %rcx,0x8(%rsp)
   1805376a5:	48 83 ec 48          	sub    $0x48,%rsp
   1805376a9:	48 c7 44 24 20 fe ff 	movq   $0xfffffffffffffffe,0x20(%rsp)
   1805376b0:	ff ff 
   1805376b2:	0f 29 74 24 30       	movaps %xmm6,0x30(%rsp)
   1805376b7:	48 8b 02             	mov    (%rdx),%rax
   1805376ba:	48 89 44 24 50       	mov    %rax,0x50(%rsp)
   1805376bf:	8b 48 f0             	mov    -0x10(%rax),%ecx
   1805376c2:	f7 c1 00 00 00 30    	test   $0x30000000,%ecx
   1805376c8:	75 0a                	jne    0x1805376d4
   1805376ca:	b9 01 00 00 00       	mov    $0x1,%ecx
   1805376cf:	f0 0f c1 48 f0       	lock xadd %ecx,-0x10(%rax)
   1805376d4:	48 8d 44 24 50       	lea    0x50(%rsp),%rax
   1805376d9:	48 89 44 24 58       	mov    %rax,0x58(%rsp)
   1805376de:	48 8d 4c 24 50       	lea    0x50(%rsp),%rcx
   1805376e3:	e8 28 35 ee 00       	call   0x18141ac10
   1805376e8:	0f 28 f0             	movaps %xmm0,%xmm6
   1805376eb:	48 8b 4c 24 50       	mov    0x50(%rsp),%rcx
   1805376f0:	48 83 c1 f0          	add    $0xfffffffffffffff0,%rcx
   1805376f4:	8b 01                	mov    (%rcx),%eax
   1805376f6:	a9 00 00 00 30       	test   $0x30000000,%eax
   1805376fb:	75 18                	jne    0x180537715
   1805376fd:	b8 ff ff ff ff       	mov    $0xffffffff,%eax
   180537702:	f0 0f c1 01          	lock xadd %eax,(%rcx)
   180537706:	ff c8                	dec    %eax
   180537708:	83 f8 ff             	cmp    $0xffffffff,%eax
   18053770b:	75 08                	jne    0x180537715
   18053770d:	e8 e2 99 bf 00       	call   0x1811310f4
   180537712:	0f 28 c6             	movaps %xmm6,%xmm0
   180537715:	0f 28 74 24 30       	movaps 0x30(%rsp),%xmm6
   18053771a:	48 83 c4 48          	add    $0x48,%rsp
   18053771e:	c3                   	ret
@@ -0,0 +1,206 @@
 soothe_mem.bin:     file format binary
 Disassembly of section .data:
 0000000000530850 <.data+0x530850>:
  530850:	48 8b c4             	mov    %rsp,%rax
  530853:	48 89 58 10          	mov    %rbx,0x10(%rax)
  530857:	48 89 70 18          	mov    %rsi,0x18(%rax)
  53085b:	57                   	push   %rdi
  53085c:	48 81 ec e0 00 00 00 	sub    $0xe0,%rsp
  530863:	0f 29 70 e8          	movaps %xmm6,-0x18(%rax)
  530867:	0f 29 78 d8          	movaps %xmm7,-0x28(%rax)
  53086b:	48 8b 05 fe 50 0e 02 	mov    0x20e50fe(%rip),%rax        # 0x2615970
  530872:	48 33 c4             	xor    %rsp,%rax
  530875:	48 89 84 24 b0 00 00 	mov    %rax,0xb0(%rsp)
  53087c:	00 
  53087d:	f3 0f 10 41 24       	movss  0x24(%rcx),%xmm0
  530882:	48 8b f9             	mov    %rcx,%rdi
  530885:	f3 0f 59 05 ff 34 f9 	mulss  0x1f934ff(%rip),%xmm0        # 0x24c3d8c
  53088c:	01 
  53088d:	8b 41 64             	mov    0x64(%rcx),%eax
  530890:	f3 0f 10 35 1c 3d f9 	movss  0x1f93d1c(%rip),%xmm6        # 0x24c45b4
  530897:	01 
  530898:	99                   	cltd
  530899:	2b c2                	sub    %edx,%eax
  53089b:	f3 0f 5e f0          	divss  %xmm0,%xmm6
  53089f:	d1 f8                	sar    $1,%eax
  5308a1:	8d 70 01             	lea    0x1(%rax),%esi
  5308a4:	66 0f 6e c6          	movd   %esi,%xmm0
  5308a8:	0f 5b c0             	cvtdq2ps %xmm0,%xmm0
  5308ab:	f3 0f 59 f0          	mulss  %xmm0,%xmm6
  5308af:	f3 0f 10 05 51 39 f9 	movss  0x1f93951(%rip),%xmm0        # 0x24c4208
  5308b6:	01 
  5308b7:	e8 f0 43 4e 01       	call   0x1a14cac
  5308bc:	ba 01 00 00 00       	mov    $0x1,%edx
  5308c1:	4c 63 c6             	movslq %esi,%r8
  5308c4:	0f 28 f8             	movaps %xmm0,%xmm7
  5308c7:	8b ca                	mov    %edx,%ecx
  5308c9:	4c 3b c2             	cmp    %rdx,%r8
  5308cc:	0f 8e 84 01 00 00    	jle    0x530a56
  5308d2:	f2 0f 10 2d 36 46 f9 	movsd  0x1f94636(%rip),%xmm5        # 0x24c4f10
  5308d9:	01 
  5308da:	49 8d 40 ff          	lea    -0x1(%r8),%rax
  5308de:	f3 0f 10 25 be 35 f9 	movss  0x1f935be(%rip),%xmm4        # 0x24c3ea4
  5308e5:	01 
  5308e6:	48 83 f8 04          	cmp    $0x4,%rax
  5308ea:	0f 8c 1c 01 00 00    	jl     0x530a0c
  5308f0:	4d 8d 50 fd          	lea    -0x3(%r8),%r10
  5308f4:	44 8d 4a 02          	lea    0x2(%rdx),%r9d
  5308f8:	0f 1f 84 00 00 00 00 	nopl   0x0(%rax,%rax,1)
  5308ff:	00 
  530900:	48 8b 87 a8 06 54 00 	mov    0x5406a8(%rdi),%rax
  530907:	0f 57 db             	xorps  %xmm3,%xmm3
  53090a:	f3 0f 2a da          	cvtsi2ss %edx,%xmm3
  53090e:	83 c2 04             	add    $0x4,%edx
  530911:	0f 28 c4             	movaps %xmm4,%xmm0
  530914:	f3 0f 5e de          	divss  %xmm6,%xmm3
  530918:	0f 28 cb             	movaps %xmm3,%xmm1
  53091b:	f3 0f 5e cf          	divss  %xmm7,%xmm1
  53091f:	f3 0f 58 cc          	addss  %xmm4,%xmm1
  530923:	f3 0f 5e c1          	divss  %xmm1,%xmm0
  530927:	0f 57 c9             	xorps  %xmm1,%xmm1
  53092a:	f3 0f 5a c8          	cvtss2sd %xmm0,%xmm1
  53092e:	0f 28 c4             	movaps %xmm4,%xmm0
  530931:	0f 54 cd             	andps  %xmm5,%xmm1
  530934:	66 0f 5a d1          	cvtpd2ps %xmm1,%xmm2
  530938:	f3 0f 59 d3          	mulss  %xmm3,%xmm2
  53093c:	0f 57 db             	xorps  %xmm3,%xmm3
  53093f:	f3 0f 11 14 88       	movss  %xmm2,(%rax,%rcx,4)
  530944:	41 8d 41 ff          	lea    -0x1(%r9),%eax
  530948:	f3 0f 2a d8          	cvtsi2ss %eax,%xmm3
  53094c:	48 8b 87 a8 06 54 00 	mov    0x5406a8(%rdi),%rax
  530953:	f3 0f 5e de          	divss  %xmm6,%xmm3
  530957:	0f 28 cb             	movaps %xmm3,%xmm1
  53095a:	f3 0f 5e cf          	divss  %xmm7,%xmm1
  53095e:	f3 0f 58 cc          	addss  %xmm4,%xmm1
  530962:	f3 0f 5e c1          	divss  %xmm1,%xmm0
  530966:	0f 57 c9             	xorps  %xmm1,%xmm1
  530969:	f3 0f 5a c8          	cvtss2sd %xmm0,%xmm1
  53096d:	0f 28 c4             	movaps %xmm4,%xmm0
  530970:	0f 54 cd             	andps  %xmm5,%xmm1
  530973:	66 0f 5a d1          	cvtpd2ps %xmm1,%xmm2
  530977:	f3 0f 59 d3          	mulss  %xmm3,%xmm2
  53097b:	0f 57 db             	xorps  %xmm3,%xmm3
  53097e:	f3 41 0f 2a d9       	cvtsi2ss %r9d,%xmm3
  530983:	f3 0f 11 54 88 04    	movss  %xmm2,0x4(%rax,%rcx,4)
  530989:	48 8b 87 a8 06 54 00 	mov    0x5406a8(%rdi),%rax
  530990:	f3 0f 5e de          	divss  %xmm6,%xmm3
  530994:	0f 28 cb             	movaps %xmm3,%xmm1
  530997:	f3 0f 5e cf          	divss  %xmm7,%xmm1
  53099b:	f3 0f 58 cc          	addss  %xmm4,%xmm1
  53099f:	f3 0f 5e c1          	divss  %xmm1,%xmm0
  5309a3:	0f 57 c9             	xorps  %xmm1,%xmm1
  5309a6:	f3 0f 5a c8          	cvtss2sd %xmm0,%xmm1
  5309aa:	0f 28 c4             	movaps %xmm4,%xmm0
  5309ad:	0f 54 cd             	andps  %xmm5,%xmm1
  5309b0:	66 0f 5a d1          	cvtpd2ps %xmm1,%xmm2
  5309b4:	f3 0f 59 d3          	mulss  %xmm3,%xmm2
  5309b8:	0f 57 db             	xorps  %xmm3,%xmm3
  5309bb:	f3 0f 11 54 88 08    	movss  %xmm2,0x8(%rax,%rcx,4)
  5309c1:	41 8d 41 01          	lea    0x1(%r9),%eax
  5309c5:	f3 0f 2a d8          	cvtsi2ss %eax,%xmm3
  5309c9:	48 8b 87 a8 06 54 00 	mov    0x5406a8(%rdi),%rax
  5309d0:	41 83 c1 04          	add    $0x4,%r9d
  5309d4:	f3 0f 5e de          	divss  %xmm6,%xmm3
  5309d8:	0f 28 cb             	movaps %xmm3,%xmm1
  5309db:	f3 0f 5e cf          	divss  %xmm7,%xmm1
  5309df:	f3 0f 58 cc          	addss  %xmm4,%xmm1
  5309e3:	f3 0f 5e c1          	divss  %xmm1,%xmm0
  5309e7:	0f 57 c9             	xorps  %xmm1,%xmm1
  5309ea:	f3 0f 5a c8          	cvtss2sd %xmm0,%xmm1
  5309ee:	0f 54 cd             	andps  %xmm5,%xmm1
  5309f1:	66 0f 5a d1          	cvtpd2ps %xmm1,%xmm2
  5309f5:	f3 0f 59 d3          	mulss  %xmm3,%xmm2
  5309f9:	f3 0f 11 54 88 0c    	movss  %xmm2,0xc(%rax,%rcx,4)
  5309ff:	48 83 c1 04          	add    $0x4,%rcx
  530a03:	49 3b ca             	cmp    %r10,%rcx
  530a06:	0f 8c f4 fe ff ff    	jl     0x530900
  530a0c:	49 3b c8             	cmp    %r8,%rcx
  530a0f:	7d 45                	jge    0x530a56
  530a11:	48 8b 87 a8 06 54 00 	mov    0x5406a8(%rdi),%rax
  530a18:	0f 57 db             	xorps  %xmm3,%xmm3
  530a1b:	f3 0f 2a da          	cvtsi2ss %edx,%xmm3
  530a1f:	ff c2                	inc    %edx
  530a21:	0f 28 c4             	movaps %xmm4,%xmm0
  530a24:	f3 0f 5e de          	divss  %xmm6,%xmm3
  530a28:	0f 28 cb             	movaps %xmm3,%xmm1
  530a2b:	f3 0f 5e cf          	divss  %xmm7,%xmm1
  530a2f:	f3 0f 58 cc          	addss  %xmm4,%xmm1
  530a33:	f3 0f 5e c1          	divss  %xmm1,%xmm0
  530a37:	0f 57 c9             	xorps  %xmm1,%xmm1
  530a3a:	f3 0f 5a c8          	cvtss2sd %xmm0,%xmm1
  530a3e:	0f 54 cd             	andps  %xmm5,%xmm1
  530a41:	66 0f 5a d1          	cvtpd2ps %xmm1,%xmm2
  530a45:	f3 0f 59 d3          	mulss  %xmm3,%xmm2
  530a49:	f3 0f 11 14 88       	movss  %xmm2,(%rax,%rcx,4)
  530a4e:	48 ff c1             	inc    %rcx
  530a51:	49 3b c8             	cmp    %r8,%rcx
  530a54:	7c bb                	jl     0x530a11
  530a56:	48 8b 9f a8 06 54 00 	mov    0x5406a8(%rdi),%rbx
  530a5d:	48 8d 15 54 03 12 02 	lea    0x2120354(%rip),%rdx        # 0x2650db8
  530a64:	48 8d 0d 4d 03 12 02 	lea    0x212034d(%rip),%rcx        # 0x2650db8
  530a6b:	ff 15 97 a5 67 01    	call   *0x167a597(%rip)        # 0x1bab008
  530a71:	44 8b ce             	mov    %esi,%r9d
  530a74:	4c 8b c3             	mov    %rbx,%r8
  530a77:	48 8b cb             	mov    %rbx,%rcx
  530a7a:	85 c0                	test   %eax,%eax
  530a7c:	75 0f                	jne    0x530a8d
  530a7e:	f3 0f 10 0d ca 33 f9 	movss  0x1f933ca(%rip),%xmm1        # 0x24c3e50
  530a85:	01 
  530a86:	e8 b5 ff c0 ff       	call   0x140a40
  530a8b:	eb 0d                	jmp    0x530a9a
  530a8d:	f2 0f 10 0d 7b 36 f9 	movsd  0x1f9367b(%rip),%xmm1        # 0x24c4110
  530a94:	01 
  530a95:	e8 66 00 c1 ff       	call   0x140b00
  530a9a:	f3 0f 10 4f 24       	movss  0x24(%rdi),%xmm1
  530a9f:	48 8d 4c 24 30       	lea    0x30(%rsp),%rcx
  530aa4:	f3 0f 10 05 24 38 f9 	movss  0x1f93824(%rip),%xmm0        # 0x24c42d0
  530aab:	01 
  530aac:	f2 0f 10 1d 8c 36 f9 	movsd  0x1f9368c(%rip),%xmm3        # 0x24c4140
  530ab3:	01 
  530ab4:	f2 0f 10 15 c4 38 f9 	movsd  0x1f938c4(%rip),%xmm2        # 0x24c4380
  530abb:	01 
  530abc:	48 8b 9f 18 07 54 00 	mov    0x540718(%rdi),%rbx
  530ac3:	0f 5a c9             	cvtps2pd %xmm1,%xmm1
  530ac6:	f3 0f 11 44 24 20    	movss  %xmm0,0x20(%rsp)
  530acc:	e8 ef 33 00 00       	call   0x533ec0
  530ad1:	48 8b 97 f8 06 54 00 	mov    0x5406f8(%rdi),%rdx
  530ad8:	4c 8d 44 24 30       	lea    0x30(%rsp),%r8
  530add:	48 8b 8f 08 07 54 00 	mov    0x540708(%rdi),%rcx
  530ae4:	4c 8b cb             	mov    %rbx,%r9
  530ae7:	89 74 24 20          	mov    %esi,0x20(%rsp)
  530aeb:	e8 90 4d 00 00       	call   0x535880
  530af0:	48 8b 97 08 07 54 00 	mov    0x540708(%rdi),%rdx
  530af7:	44 8b c6             	mov    %esi,%r8d
  530afa:	48 8b 8f f8 06 54 00 	mov    0x5406f8(%rdi),%rcx
  530b01:	e8 ca 4a 00 00       	call   0x5355d0
  530b06:	48 8b 97 f8 06 54 00 	mov    0x5406f8(%rdi),%rdx
  530b0d:	44 8b c6             	mov    %esi,%r8d
  530b10:	48 8b 8f a8 06 54 00 	mov    0x5406a8(%rdi),%rcx
  530b17:	e8 74 ce ff ff       	call   0x52d990
  530b1c:	48 8b 87 a8 06 54 00 	mov    0x5406a8(%rdi),%rax
  530b23:	c7 00 00 00 00 00    	movl   $0x0,(%rax)
  530b29:	48 8b 8c 24 b0 00 00 	mov    0xb0(%rsp),%rcx
  530b30:	00 
  530b31:	48 33 cc             	xor    %rsp,%rcx
  530b34:	e8 97 05 c0 00       	call   0x11310d0
  530b39:	4c 8d 9c 24 e0 00 00 	lea    0xe0(%rsp),%r11
  530b40:	00 
  530b41:	49 8b 5b 18          	mov    0x18(%r11),%rbx
  530b45:	49 8b 73 20          	mov    0x20(%r11),%rsi
  530b49:	41 0f 28 73 f0       	movaps -0x10(%r11),%xmm6
  530b4e:	41 0f 28 7b e0       	movaps -0x20(%r11),%xmm7
  530b53:	49 8b e3             	mov    %r11,%rsp
  530b56:	5f                   	pop    %rdi
  530b57:	c3                   	ret
  530b58:	cc                   	int3
  530b59:	cc                   	int3
  530b5a:	cc                   	int3
  530b5b:	cc                   	int3
  530b5c:	cc                   	int3
  530b5d:	cc                   	int3
  530b5e:	cc                   	int3
  530b5f:	cc                   	int3
--- a/Show More
+++ b/Show More