docs: index frank-primary-freetoken-pr587; mark the r9v q8 runbook retired

All ~173 managed skills read in full and merged, one polished reference
page per skill-cluster:

- LLM cross-cutting (17 pages): GGUF Compatibility, Llama Cpp Serving
  Sizing, Llama Cpp Multi-Device Splits, PCIe Link and Atomics, RDNA4
  Container Serving, RDNA4 MoE Expert Offload VRAM Budget, vLLM Boot
  Forensics, vLLM Porting and Compile Triage, LLM Bench Windows, LLM
  Performance Diagnosis, Model Swap and Rollback, Tcclaviger Fork Ops,
  Heterogeneous Split Triage, FreeToken Builds + deltas folded into the
  retired runbooks (GLM DFlash2 port laws, Lucebox dual-R9700, R9V
  qualification pipeline, FLM pack registration, ik hybrid envelope,
  frank-medium split rebalance, llama.cpp GGUF preflight, batch-ubatch
  split section, bench-window D-state/stale-read notes, second-endpoint
  bring-up).
- Architecture (3): Stale Signal Audit, Quadlet Lifecycle, Deploy State
  Triage.
- Operations (6 + index): Remote Job Runner, Payload and Disk Reclaim,
  Container Host Reachability, Fleet Ops Playbook, Fleet Audit and
  Posture, SSH and Remote Access.
- Infrastructure: Observability Provisioning + Homepage Status Probes;
  updates to Authentik, Anubis, LAN-DNS (per-client DoH bypass, Porkbun
  zone ops), Swap Configuration, OpenWebUI, Hermes (pre-existing page
  preserved and extended).
- FrankBook-Pro: ARR-Stack filled (API split, schema-first probing,
  auth-posture, auth-config trap); qBittorrent E2E verification.
- Desktop/Agent/Research: Noctalia and Niri, OMP Harness, Agent Docs and
  Edits, Command Center Dotfiles, PICO-8 Asset Standards.
- Indexes: LLM Cross-Cutting References section, Operations/ and Agent/
  index pages, root-index Skill Distillation section, AGENTS.md pointer.
- log.md carries the wave plan.

check-vault-links.py 0 unresolved (185 pages); ./tests/run.sh 366/366.

AGENTS.md

@@ -12,8 +12,8 @@ merge absorbed the former `vllmScripts` repo, history preserved).
   Calibre-Web, Jellyfin, Plex) plus the `frank-primary` LLM service.
 - **Dankclaw** (192.168.0.143) — AI/search/dev/observability stack (SearXNG
   via Gluetun, ComfyUI, OpenWebUI, Hermes, Gitea, Grafana, Prometheus, Loki,
-  Alloy, Pi-hole) plus TWO LLM services: `frank-large` (stock llama.cpp on
-  the gfx1151 iGPU) and `frank-small` (llama.cpp on the RTX 3090). Caddy on
+  Alloy, Pi-hole) plus TWO LLM services: `frank-large` (halogen container on
+  the gfx1151 iGPU) and `frank-small` (llama.cpp CUDA on the RTX 3090). Caddy on
   Dankclaw reverse-proxies both hosts.
 - **FrankBook-Pro** (192.168.0.112) — macOS mini running the *arr stack
   (Prowlarr, Sonarr, Radarr, Lidarr, Readarr, qBittorrent) as quadlets inside
@@ -24,9 +24,9 @@ merge absorbed the former `vllmScripts` repo, history preserved).
 
 | Target | Service | Runtime | Model | Port |
 |--------|---------|---------|-------|------|
-| Dankserver | `frank-primary` | R9V radiance-vLLM fork `drwolfen/radiance-vllm-qwen4exp` (docker-pinned image bundle `2dac17a2`, docker→podman shim, root systemd unit) | unsloth Qwen3.8-Flash-Next-GGUF Q8_0 (175.30 GiB, 6 GGUF shards) with the blockfp8 MTP4 head (depth 4) and mmproj, TP=2 tiered experts (80-slot LRU cache; PLE n-gram PINNED in host RAM 54.4 GB; quantized inject-shard dev-overlay adapter), CED off, 131072 ctx single stream, Qwen-Sharp template, tool parser `qwen3_coder`, reasoning channel `reasoning` | 7770 |
-| Dankclaw | `frank-large` | unsloth llama.cpp b11030 prebuilt (ROCm gfx1151, root systemd unit) | unsloth Qwen3.8-27B-GGUF:UD-Q8_K_XL (27B dense, 31.5 GiB) with the z-lab DFlash2 Q4_K_M drafter (depth 4) on the gfx1151 iGPU (Strix Halo LPDDR, 114 GiB GTT), 262144 ctx single stream (1 slot), q8_0 KV, text only | 7770 |
-| Dankclaw | `frank-small` | mainline llama.cpp master 37b53fd (root systemd unit) | HauhauCS Gemma4-12B-QAT-Uncensored-Balanced Q4_K_M (BF16 mmproj + MTP assistant sidecar, `draft-mtp,ngram-mod`) on the RTX 3090, 1048576 pool / 4 slots of 262,144, q8_0 KV, the fleet's fastest vision endpoint | 7771 |
+| Dankserver | `frank-primary` | **FreeToken PR #587** (branch `rocm714_torch214`, head `77a7201`; local image `frank-primary-freetoken:pr587-77a7201` built on the host from `scripts/docker/frank-primary-freetoken.Dockerfile` — digest-pinned rocm/pytorch rocm7.14 torch-2.11.0 base, plain `podman run`, root systemd unit) | `orcarouter/Qwen3.8-Flash-Next-Uncensored-FP8` (~172.8 GiB, 131 shards, modelopt block-FP8, loaded straight from safetensors), **ONE R9700 (second card idle)**, all experts in **pinned host RAM** (~120 GiB) behind a GPU LRU (`--moe-cache-rate 0.10`), PLE table 47.7 GiB pinned (root unit + `LimitMEMLOCK=infinity`), **KV pool `--num-tokens 299968`** (24 KiB/token, multiple-of-64 law of the `qsa_sparse` auto backend), 2 slots, `--max-output-tokens 36864`, **parsers AUTO** (thinking → `reasoning_content`, tools qwen3.8 dialect), vision tower host-streamed. Measured: boot ~2.3 min cold / 92 s warm, decode **17.1 tok/s** single-stream; **TP=2 does not boot on ROCm** (pynccl JIT); `/metrics` 404 (Prometheus scrape reads down). Deploy gate reads the `/health` BODY (`status:"ok"`) — FreeToken 200s while loading. Rollback tag `frank-primary-vllm-shapeG-20261003` |7770|
+| Dankclaw | `frank-large` | **halogen container** (`ghcr.io/peonist-ai/halogen@sha256:dc0a39a0…`, digest-pinned, plain `podman run` via `scripts/serve-frank-large.sh`, root systemd unit) | `peonist-ai/halogen-qwen3.8-27b` **`qwen3.8-27b-p1w4d-d2.hgn`** (35,865,565,184 B; the `.hgn` format loads in NO other engine — llama.cpp, vLLM and transformers all refuse it), **DFlash2 spec decode (`HALOGEN_DRAFTER=2`)**, **1 slot × 200,700 ctx** — `HALOGEN_KV_SLOTS=1` is load-bearing: batching XOR speculation, slots>1 silently disables the drafter (measured: 8 slots at serial ~6 t/s/stream), d2 KV (2× fp8, 64 KiB/token → 12.2 GiB pool, engine sizes it and prints a boot ledger naming any clamp), **TEXT-ONLY** (the engine ignores the vision encoder — vision is frank-small's job), **censored base** (the ROCmFPX AGENT rollback is the refusal escape hatch), **no server-side sampler: absent temperature is GREEDY**, so the card's thinking recipe (temp 1.0 / top_p 0.95) rides the wire in `models.yml`. Publisher card 31.7 t/s mean across eight text workloads vs the retired ROCmFPX endpoint's measured 27.6–28.1. Prompt over the slot AND `max_tokens` over `HALOGEN_MAX_TOKENS_CAP` are both 400s naming the limit, never truncations. Serve passes NO `--ulimit nofile` (rootless runc cannot raise Dankclaw's 524288 hard limit — see Key Lessons). Cutover 2026-10-05|7770|
+| Dankclaw | `frank-small` | llama.cpp master, CUDA sm_86 build (root systemd unit, relinked) | bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF **Q6_K** (9B dense qwen3_5 hybrid, 7.26 GiB GGUF, no MTP weights so no spec decode) + the repo's f16 mmproj for **vision**, on the RTX 3090, **2 × 200,704** (pool 401,408 — NOT the native window) with **fully unquantized KV (f16 K + f16 V)** under flash attention, **batch 4096 / ubatch 2048** (doubled 2026-10-05: measured +4.6 % prefill at 8,928 tokens for 616 MiB), 21,846 used / **2,279 free** of 24,576 MiB, 94.4 tok/s decode. Resized 2026-10-03 (3 × 180,224 + q8_0 V → 2 × 200,704 + f16 V → + Q6_K rung: the freed slots paid for exact values AND the higher quant; the rung costs a measured −10 % decode — weight-bandwidth-bound); rollback `MODEL_FILE=…/MiMo-Q5KM/…Q5_K_M.gguf` (buys the 10 % back) or `CTX_SIZE=540672 PARALLEL=3 CACHE_TYPE_V=q8_0` (Q5 only) | 7771 |
 
 Retired endpoints (`frank-npu`, and every prior runtime) are in
 the [runbook index](#runbook-index). The full swap history lives in
@@ -40,7 +40,11 @@ the [runbook index](#runbook-index). The full swap history lives in
 - `scripts/` — `deploy.sh` / `deploy-plan.py` / `deploy-remote.sh`, LLM serve
   scripts, systemd units, env tunables, bench harnesses
 - `old/` — retired LLM scripts (reference only)
-- `serverAppConfigs-docs/` — knowledge base (Obsidian vault, OKF format)
+- `serverAppConfigs-docs/` — knowledge base (Obsidian vault, OKF format).
+  The managed skill library is distilled there one page per skill-cluster:
+  `LLM/index.md` → *Cross-Cutting References*, plus `Operations/index.md`,
+  `Agent/index.md`, and the Infrastructure/Dankclaw/FrankBook-Pro service
+  pages (`serverAppConfigs-docs/log.md` carries the wave plan and results)
 - `tests/` — bats suite + Playwright e2e
 
 ## Git Workflow
@@ -74,6 +78,14 @@ unrelated
 container changes cannot interrupt inference. Set `FRANK_SUDO_PW` for headless
 runs on a host without passwordless sudo; the value never enters the repo.
 
+**`deploy.sh <target>` with no `--service` is a fleet-wide DRAIN, not a sync.**
+It also walks every per-service revision marker under the deploy-state dir and
+restarts each service whose marker is behind HEAD *and* whose paths changed
+since that marker — so a backlog accumulated by service-scoped deploys fires
+all at once, and one failing health gate aborts the run with every later
+service left pending. Name the service unless a drain is what you want
+(`./scripts/deploy.sh dankserver --service frank-primary`).
+
 ### Deployment completeness (MANDATORY)
 
 A deploy is **not complete** when `deploy.sh` exits green. It requires:
@@ -82,7 +94,14 @@ A deploy is **not complete** when `deploy.sh` exits green. It requires:
 2. the Playwright e2e suite (`tests/e2e/run.sh`) — runs **automatically** as
    the `[3/3]` gate at the end of `deploy.sh` against the LIVE fleet; a
    behavior-affecting deploy that fails it exits `DEPLOY INCOMPLETE`. Bypass
-   only with `--skip-e2e` in an emergency.
+   only with `--skip-e2e` in an emergency. The gate runs the conditional
+   `frankbook` Playwright project (the `*arr`/qBittorrent walk,
+   `tests/e2e/specs/19-frankbook.spec.ts`) ONLY when a frankbook blast-radius
+   path changed since origin/main — that spec, `frankbook-pro/**`,
+   `shared/homepage/services.json`, or the `dankclaw/Caddyfile` serving the
+   arr vhosts — so a sleeping FrankBook-Pro never fails an unrelated deploy.
+   The default project set excludes it everywhere else (`E2E_INCLUDE_FRANKBOOK=1`
+   opts in; see `serverAppConfigs-docs/Architecture/E2E Suite.md`).
 
 ```mermaid
 graph LR
@@ -208,13 +227,27 @@ archive.
 
 **STOP the production service BEFORE benchmarking** — these servers serve
 real users, and a bench beside a live service starves the GPU and corrupts
-results. Use the runtime's own harness: `llama-bench` for every llama.cpp
-endpoint (all three are llama.cpp now). Rolling your own bench
+results. Use the runtime's own harness: `llama-bench` for the llama.cpp
+endpoints; a live llama.cpp server also reports honest per-request numbers in the `/completion` `timings` object (prefill/decode rates, `draft_n`, `draft_n_accepted`), which is how frank-large and frank-small were measured. Rolling your own bench
 requires asking first.
 Scratch ports are not used on LLM hosts — a
 test process cannot share the single GPU anyway; validate against the service
 itself.
 
+(A gufo-era law; frank-large ran ROCmFPX llama.cpp 2026-10-02 → 10-05 and is the halogen container since, but the rollback tags keep it live.) **`gufo bench` cannot rank speculative-decode settings, and will invert the ranking.**
+Measured 2026-09-28 on frank-large at depth 32768: its corpus
+reports `acceptance=0.933 ± 0.000` with an **identical trace sha256** across
+draft widths, i.e. the same highly-predictable text every run, and 93 % of
+every proposal accepted. It reported 29–37 t/s decode where real traffic
+gives ~15 t/s, and it made `--draft-tokens 16` look 6 % faster than 7 when
+real text made it 14 % *slower*. Use it for shape, not for A/B. For an A/B,
+replay **real** text through the live server and read the `Server-Timing`
+header — but note the unit rides in the **name** half (`ttft;dur=402.485`),
+so splitting on `=` alone keys the dict `ttft;dur` and every lookup silently
+returns nothing. Generate **at least 128 tokens** for the measurement: at 8
+tokens, `inter_token` is first-token noise and the run-to-run spread swamps
+every real difference.
+
 This covers **any GPU experiment, not just benchmarks** — including a bare
 2-rank RCCL/`torch.distributed` probe. On 2026-09-16 one such probe was run
 against the live service "because it only needs 200 MB of VRAM"; the host went
@@ -227,11 +260,11 @@ one: `ip neigh show <ip>` from a peer — FAILED means off/NIC dead, while a
 merely wedged kernel usually still answers ARP.
 
 1. Stop and mask (self-healing would restart it):
-   - Dankclaw `frank-large` (root unit, stock llama.cpp on the iGPU):
+   - Dankclaw `frank-large` (root unit, halogen podman on the iGPU):
      `sudo systemctl stop frank-large && sudo systemctl mask frank-large`
    - Dankclaw `frank-small` (root unit, llama.cpp on the 3090):
      `sudo systemctl stop frank-small && sudo systemctl mask frank-small`
-   - Dankserver `frank-primary` (root unit, R9V radiance on the 2x R9700):
+   - Dankserver `frank-primary` (root unit, FreeToken on one R9700):
      `sudo systemctl stop frank-primary && sudo systemctl mask frank-primary`
 2. Bench.
    NOTE: `systemctl mask` FAILS on a relinked unit (the unit file in
@@ -263,168 +296,216 @@ When renaming a service, retiring an endpoint, or moving a model: update the
 provider block, model `id`, `contextWindow`, `maxTokens`, `input` in
 `models.yml`, then fix every `modelRoles` entry that named the old endpoint —
 a role pointing at a provider the file does not define fails at call time. As
-of 2026-09-26 both `frank-small` and `frank-primary` accept images
-(`input: [text, image]`) — frank-primary loads the BF16 mmproj explicitly
-(this build never autoloads projectors); `frank-large` remains text-only
-because its stock llama.cpp deployment loads no
-vision projector (the mmproj is not loaded). Note also that
-the reasoning channel differs by runtime: llama.cpp endpoints emit
-`reasoning_content`, vLLM emits `reasoning` — academic since the
-2026-09-26 swap, as every endpoint in the fleet is llama.cpp now.
+of 2026-10-05 **two of the three endpoints accept images** (`input: [text,
+image]`) — `frank-large` went TEXT-ONLY with the halogen swap (the engine
+ignores the vision encoder; the BF16-mmproj era ended 2026-10-05) — while
+`frank-primary` takes them natively (FreeToken builds the qwen3.8 vision
+tower and streams its block weights from host RAM two at a time by default),
+and `frank-small` through
+`mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf` passed as `--mmproj` — the MiMo
+GGUF itself carries no vision tensors. `frank-small`'s `contextWindow` is
+**200704** (resliced from 180224 on 2026-10-03), the per-slot window, not the
+model's native 262144: the harness must compact before llama.cpp answers a
+longer prompt with a 400. Note also
+that
+the reasoning channel differs by runtime: llama.cpp AND SGLang (qwen3
+parser, live-verified) emit `reasoning_content`, vLLM emits `reasoning` —
+models.yml sets `reasoningContentField` per endpoint to match.
 There is **no audio**
-input anywhere in the fleet except frank-small: that capability left with the
-FLM NPU endpoint on 2026-09-11 and returned on 2026-09-17 with the Gemma4
-projector's audio encoder (frank-small transcribes WAV verbatim even though
-`audio` is omitted from its `input` list — OMP's schema rejects it).
+input anywhere in the fleet: that capability left with the FLM NPU
+endpoint on 2026-09-11, returned 2026-09-17 with the Gemma4 projector's
+audio encoder, and left again with the Gemma4 retirement on 2026-10-01
+(the MiMo tower is vision-only).
 
-### frank-large — stock llama.cpp Qwen3.8-27B UD-Q8_K_XL (Dankclaw, ACTIVE)
+### frank-large — halogen container, Qwen3.8-27B + DFlash2 (Dankclaw, ACTIVE)
 
-The unsloth llama.cpp prebuilt `app-b11030-mix-5ff778e` (ROCm gfx1151,
-`/home/dank/GitPackages/unsloth-llama-b11030/`, bundles its own ROCm
-runtime — no host SDK gate) as a **root systemd unit**
-(`scripts/frank-large.service`, `User=dank`), serving
-`unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL` (27B dense, 31.5 GiB,
-sha256-verified) on port 7770 from the gfx1151 iGPU. One stream, full
-262,144 ctx (the model's native window, `PARALLEL_SLOTS=1`), DFlash2 spec
-decode (the z-lab `Qwen3.8-27B-DFlash2-Q4_K_M` drafter, depth 4,
-`SPEC_DRAFT_N_MAX=4`), reasoning extracted to `reasoning_content`, and
-**no vision projector**. It replaced the Kairic Edge ROCmFPX build (jcbtc
-Qwen3.8-27B-IU4-Kairic-Edge), which had replaced the unsloth
-Qwen3.8-Flash-Next UD-IQ4_XS.
+The **halogen engine** (`ghcr.io/peonist-ai/halogen@sha256:dc0a39a0…`,
+digest-pinned, published tag 0.1.4) as a podman container driven by
+`scripts/serve-frank-large.sh` under a **root systemd unit**
+(`scripts/frank-large.service`, `User=dank`, `Restart=always`, `ExecStop=podman
+stop -t 120 frank-large-halogen`, the script `exec podman wait` so the restart
+policy is real) on port 7770 from the Strix Halo iGPU. Engine stdout goes to
+`/home/dank/logs/frank_large.log` — **the boot ledger (KV pool, drafter state,
+any clamp) lives there, not in the journal**. Serving
+`peonist-ai/halogen-qwen3.8-27b`: `qwen3.8-27b-p1w4d-d2.hgn` (35,865,565,184 B
+≈ 33.4 GiB, sha256 `274c3fc7…`, 1352 tensors; p1 weights / w4 activations /
+d2 KV) + the flat `tokenizer/` directory (5 files). Cutover 2026-10-05 from the ROCmFPX llama.cpp fork
+([[frank-large-rocmfpx-q6agent-uncensored-mtp]], measured 27.6–28.1 t/s prose),
+chosen for the publisher's 31.7 t/s DFlash2 mean.
 
-The spec method is DFlash2 (block-diffusion drafter) at depth 4. The
-drafter is loaded via `--model-draft`; the chain is `--spec-type
-draft-dflash` with `--spec-draft-device ROCm0`. KV is **q8_0** (the
-unsloth b11030 build does not support fp8 KV; q8_0 is the
-highest-precision 8-bit option, ~8 GiB at 262k). The serve script
-(`scripts/serve-frank-large.sh`) validates binary → main GGUF → drafter
-and exits with a pointer message if any is missing, then `exec`s
-`llama-server` with `--jinja --alias frank-large --chat-template-kwargs
-'{"preserve_thinking":true}' --chat-template-file $TEMPLATE_FILE` (the
-Qwen-Sharp v22.4.0 template; without `--chat-template-file` the binary
-falls back to the built-in `peg-native` template, which produces a
-~73,600-token prompt for a ~250-token request).
+**The format is the lock.** `.hgn` loads in nothing else — the runtime is
+chosen by the FILE, as ROCmFPX's tensor types were before it. This is NOT the
+2026-09 `halogen-flash-server` image (different repo, Flash-Next w4b era); the
+halogen-server repo restarted versioning at 0.1.x.
 
-The iGPU addresses **114 GiB** (the GTT aperture over the shared LPDDR,
-`gtt_total` in `/sys/class/drm/card1/device/`; `rocm-smi`'s 2 GiB "dedicated"
-is not the budget). The 27B dense model (31.5 GiB) is mmap'd in with
-`--n-gpu-layers 999`; budget at 262k ≈ 41-43 GiB (weights + KV + buffers),
-leaving ~71 GiB of aperture headroom. The 8 GiB prompt-cache budget
-(`CACHE_RAM=8192`) is the qualified config.
+**Two product regressions, both accepted by Frank before deploy:** the
+endpoint is **text-only** (README honest limits: the vision encoder is
+ignored — frank-small owns images now) and the checkpoint is a finetune of the
+**censored** `Qwen/Qwen3.8-27B` base (refusals can return; the rollback is one
+commit-revert away with the AGENT payload still on disk).
 
-Fetch: `scripts/download-frank-large.sh` (per-file, each file verified
-against its declared byte count + sha256; `--verify-only` re-checks).
+**Engine laws that shape the config** (all in `scripts/frank-large.env` with
+evidence): `HALOGEN_KV_SLOTS=1` — batch XOR speculation, slots>1 turns the
+drafter off silently. `HALOGEN_SLOT_CTX=200700` — the per-conversation HARD
+ceiling (400 naming the limit, never truncation); `models.yml`
+`contextWindow: 200700` must match or the harness discovers the wall as a 400.
+`HALOGEN_MAX_TOKENS_CAP=65536` + `HALOGEN_QUEUE_TIMEOUT=7200` are a coupled
+pair; the harness `maxTokens: 57344` stays under the cap (over-cap is a 400
+naming the limit — README, never a 413). Thinking tokens COUNT against the
+budget: exhausted mid-thought is `finish_reason: "length"` with empty
+`content` and the reasoning in `reasoning_content`.
+`HALOGEN_DRAFTER=2` = DFlash2. **No server-side sampler**: an absent
+temperature is greedy, so the thinking recipe exists only in `models.yml`
+`extraBody`. `response_format` is unsupported (no constrained decoding).
+Thinking lands in **`reasoning_content`** (harness field unchanged); the same
+Qwen3.8 template rejects `reasoning_effort` outside `xhigh|medium|low`, so the
+`models.yml` fold stays mandatory.
 
-Runbook: `serverAppConfigs-docs/LLM/frank-large-unsloth-qwen38-27b-q8xl.md`.
-Rollback: [[frank-large-kairic-edge-qwen38-27b]] (payload and runtime
-intact on disk).
+Fetch: `scripts/download-frank-large-halogen.sh` into
+`/home/dank/models/halogen-qwen3.8-27b/` (`-f` + byte count + LFS sha256 +
+`--verify-only`, no token on the command line; resume goes against the
+RESOLVED CDN URL because curl drops `Range` across HF's 302). The deployer
+and serve script both refuse a missing tokenizer file or **any symlink** —
+an HF cache snapshot symlinks each entry into `blobs/` and those links
+dangle inside the container mount.
+
+Runbook: `serverAppConfigs-docs/LLM/frank-large-halogen27b-dflash2.md`.
+Rollback: revert the swap commit and redeploy — the ROCmFPX payload
+(`/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP/`), the fork build
+and `scripts/build-rocmfpx.sh` (pin `36e9acd`) all remain on disk, seam-tested.
+The older gufo tag `frank-large-gufo-q4kxl-20261001` stays a **re-download**
+rollback (payloads removed 2026-10-03; `scripts/build-gufo.sh` stays).
 
 ### frank-small — llama.cpp on the RTX 3090 (Dankclaw, ACTIVE)
 
-Mainline llama.cpp master `37b53fd` (CUDA build with
-`-DGGML_CUDA_FA_ALL_QUANTS=ON`) as a **root systemd unit** (repo
-`scripts/frank-small.service`, `relink_unit`ed into `/etc/systemd/system/`),
-serving `HauhauCS/Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced` at Q4_K_M.
-The MTP head is a SEPARATE assistant file (`mtp-gemma-4-12B-it.gguf`, 4
-nextn layers) loaded via `--model-draft`; the draftless **ngram-mod**
-matcher chains behind it in the same `--spec-type` value. BF16 mmproj,
-port 7771, q8_0 KV, `CTX_SIZE=1048576` / `PARALLEL=4`
-(four FULL 262,144-token slots — the measured card ceiling), batch
-**8192/1024**, greedy default. Deployed flags are in
-`scripts/frank-small.env`; the serve script refuses to start if the
-weights, the projector, the sidecar or the binary are missing.
+**llama.cpp master** (CUDA sm_86 build
+`/home/dank/GitPackages/llama-master/build-cuda/bin/llama-server`) as a
+**root systemd unit** (`scripts/frank-small.service`, `relink_unit`ed into
+`/etc/systemd/system/`), serving
+`bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF` at **Q6_K** (7,793,710,624 B,
+sha256 `ef96d05a…`) **plus `mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf`**
+(918,166,048 B, sha256 `ff348f…`), both `/home/dank/models/MiMo-Q6K/` (the
+projector is a hardlink of the Q5KM copy — the dir is self-contained), on
+port 7771. The rung went Q5_K_M → Q6_K on 2026-10-03 the same morning as the
+reslice, funded by the freed slots. The runtime arrived 2026-10-01 night for
+one reason:
+**asymmetric KV** — `--cache-type-k f16` + `--cache-type-v q8_0` is the fleet's
+only per-tensor KV lever, and SGLang's `--kv-cache-dtype` moves K and V
+together. On 2026-10-03 Frank spent the headroom that lever bought: the reslice
+from three slots to two made room for **fully unquantized V**
+(`CACHE_TYPE_V=f16`); both flags stay explicit in the serve script. Vision came
+2026-10-02 via `--mmproj`: the GGUF carries no `v.*` tensors, so the projector
+file is the entire vision path (and ~1.4 GiB of the card).
 
-Verified at deploy (2026-09-17): `/v1/models` reports `frank-small`; a
-chat completion returns content plus `reasoning_content`; a tool call
-returns `tool_calls`; an image request is answered from the projector;
-an `input_audio` WAV transcribes verbatim. Post-tuning live numbers at
-max length: pp@254k ~965 t/s, tg@253k 226–302 t/s with ~94% draft
-acceptance; idle 23,300 MiB of 24,576. The throughput sweep (and why
-ubatch 2048 OOMs the draft's pp buffer, and why draft depth 4 + poll
-100 is WORSE at long ctx) is tabled in the runbook.
-Runbook: `serverAppConfigs-docs/LLM/frank-small-gemma4-12b.md`.
-Rollback: [[frank-small-qwen38-gsq-rco]].
+**Architecture: qwen3_5 hybrid** — 32 layers, 8 full-attention + 24 Gated
+DeltaNet, native 262,144 window. KV exists only in the 8 attention layers
+(4 KV heads × 256 dim × 2 B ≈ 2 KiB/token/layer, ~16 KiB/token for f16 K+V);
+the DeltaNet state is a fixed per-slot buffer, so slot count — not context —
+drives that term.
 
-### frank-primary — R9V radiance-vLLM Qwen3.8-Flash-Next Q8_0 (Dankserver, ACTIVE)
+**The shape (2026-10-03):** `--ctx-size 401408 --parallel 2` = **TWO
+conversations of 200,704 tokens each** (llama.cpp divides the pool:
+`n_ctx_slot = CTX_SIZE / PARALLEL`), `--flash-attn on` (the seam pins it),
+**f16 K + f16 V**, all layers on the GPU, `--mmproj` for vision, and since
+2026-10-05 **batch 4096 / ubatch 2048** (doubled; the boot-pair A/B is in the
+runbook). Measured trade at 8,928 prompt tokens: prefill 4,024 → **4,209 t/s,
++4.6 %**, decode flat (91.9–93.0 vs 93.8–94.3 — noise), and a 279-token prompt,
+which sits below either ubatch, did not move at all. The old "ubatch 2048
+failed the scratch boot" note belongs to the Q5_K_M predecessor and does NOT
+generalize: this shape boots 2048 clean. Measured VRAM **21,846 used / 2,279
+free** of 24,576 MiB against **21,276 / 2,849** for the 1024 arm, reproduced
+byte-identically the same hour — the doubling costs **616 MiB** and lands at
+the BOTTOM of the 2–3 GiB floor Frank asked for (`4096/1536` is the middle rung
+if that margin is ever wanted; non-KV 9,302 MiB, up from 8,732 by exactly the
+compute buffers; f16 K+V is 32 KiB/token across the 8 attention layers;
+quote `nvidia-smi`'s `free` field, never `total − used` — the driver holds
+~450 MiB of reserve the subtraction hides). The 2026-10-02 3 × 180,224 +
+q8_0-V shape stays one env edit away (`CTX_SIZE=540672 PARALLEL=3
+CACHE_TYPE_V=q8_0`, and it must pair with Q5_K_M — Q6_K breaks the floor
+there). Three native windows need 18.4 GiB of KV alone and were never on the
+table.
 
-The **R9V radiance-vLLM fork** (`drwolfen/radiance-vllm-qwen4exp`, docker-pinned
-prebuilt image bundle `r9v-qwen38-flash-next-wmma@sha256:2dac17a2…`) serving
-the **unsloth Qwen3.8-Flash-Next-GGUF Q8_0** (175.30 GiB, 6 GGUF shards)
-with the blockfp8 MTP4 head (depth 4) and the mmproj on **both** Radeon AI
-PRO R9700s (gfx1201, BDFs `0000:03:00.0` / `0000:07:00.0`), one stream at
-**131,072 ctx**, **TP=2 tiered experts** (80-slot LRU expert cache; PLE
-n-gram table PINNED in host RAM, 54.4 GB), the **Qwen-Sharp v22.4.0
-template** (via the launcher's `R9V_CHAT_TEMPLATE_FILE` passthrough, fork
-`30d4a32`), tool parser `qwen3_coder`, reasoning channel `reasoning`. Cut
-over 2026-09-27 from the uncensored IQ4_XS payload (itself cut over from
-llama.cpp b11030 the same day) at Frank's direction.
+Verified at deploy (2026-10-03): boot logs `n_slots = 2, n_ctx_slot = 200704`
+and `loaded multimodal model`; decode **94.44 tok/s** at temp 0 on the probe
+that measured 104.84 on Q5_K_M — **the Q6_K rung costs a measured −10 %**:
+decode on this card is weight-bandwidth-bound and Q6_K streams ~10 % more
+bytes/token; health 200, e2e 33/33; native `tool_calls`; thinking in
+`reasoning_content`; a **512×512** solid-red PNG answers `Red` through an
+`image_url` content part (two traps: the legacy top-level `images` field is
+ignored, and a TINY probe image undersupplies image tokens — a 10×10 PNG made
+the model claim "it's a video"; probe with ≥256 px). A prompt over the slot
+is a **400 `exceed_context_size_error`**, which is why `models.yml` sets
+`contextWindow: 200704`.
 
-**The tuned shape lives on-host, not in the repo.** The q8 profile's setup
-run qualified the `R9V_*` config into
-`/home/dank/r9v-state/q8xl/setup.json` (`ready: true`); the repo env
-file `scripts/frank-primary.env` carries ONLY production overrides
-(`R9V_HOST_BIND=0.0.0.0`, `R9V_HOST_PORT=7770`,
-`R9V_SERVED_MODEL_NAME=frank-primary`, `R9V_CONTAINER_NAME=frank-primary-r9v`,
-`R9V_EXPECTED_GPU_BDFS`) plus two q8-only keys: the dev-overlay adapter
-(`R9V_DEV_QWEN4_EXP_ADAPTER_PY`, fork commit `363d606` — the Q8_0 payload
-quantizes BOTH fused hyper-connection shards, so the shard-uniformity
-check needs the in-stream dequant) and `R9V_FULL_MUTABLE_CACHE=0` (the
-overlay's env.always injects the uncensored 160-slot cache geometry, which
-the q8 80-slot tiered cache rejects). Never copy tuned values from the
-state into the repo — it would drift from the qualification.
+Runbook: `serverAppConfigs-docs/LLM/frank-small-llama-mimo-q6k-2x200704.md`
+(the boot-pair table that prices pool changes lives in the retired
+[[frank-small-llama-mimo-q5km]]).
+Rollback: `MODEL_FILE=/home/dank/models/MiMo-Q5KM/MiMo-V2.6-Distill-Qwen-9B-Q5_K_M.gguf`
+buys back 875 MiB and the −10 % decode rate (3,571 free, 104.84 tok/s, both
+measured); the older shape `CTX_SIZE=540672 PARALLEL=3 CACHE_TYPE_V=q8_0`
+pairs with Q5 only; SGLang rollback in
+[[frank-small-sglang-mimo-v26]] (venv on disk at
+`/home/dank/GitPackages/sglang-cu128/`, INT8 payload staged to
+`~dank/.AGENTTRASH/20261002-retired-llm-payloads/`).
 
-**Spec decoding is MTP4 only.** The fork's `speculative_config` is a
-single dict — one method per engine, NO chaining — so the llama.cpp
-`draft-mtp,ngram-mod` pattern is NOT expressible here. MTP4 is wired
-(`R9V_MTP_SPEC_TOKENS=4`, draft TP=2); the PLE n-gram rides along
-architecturally (`ngram_embedding`); `ngram_gpu` exists as an alternative
-single method, not a chainable one.
+### frank-primary — FreeToken (PR #587 ROCm), Qwen3.8-Flash-Next FP8 (Dankserver, ACTIVE)
 
-**Serving = the fork's own launcher through a docker→podman shim.** The R9V
-tooling is docker-pinned; the shim at
-`/home/dank/GitPackages/radiance-vllm-qwen4exp/.podman-shim/docker` carries
-three laws: (1) the bundle is a containerd/OCI docker-save, so docker keys
-the image by its OCI manifest digest while podman keys by config digest but
-preserves the manifest digest in `RepoDigests` — the shim answers
-`image inspect sha256:X` from RepoDigests and rewrites bare `sha256:X` run
-refs to `repo@sha256:X`; (2) podman inspect answers
-`HostConfig.LogConfig.Config: null` where docker returns the log-opts map
-(the fork's doctor crashes on it) — the shim records `run --log-opt` pairs
-and normalizes inspect output; (3) podman lacks docker's `SecurityOptions`
-/`ServerVersion` info fields — the shim answers them.
+The **FreeToken engine at PR #587 head `77a7201`** (branch
+`rocm714_torch214`, built on the host into the local image
+`localhost/frank-primary-freetoken:pr587-77a7201` by
+`scripts/docker/frank-primary-freetoken.Dockerfile` — the PR's own
+docs/install_amd.md stack: digest-pinned
+`rocm/pytorch:rocm7.14_ubuntu24.04_py3.12_pytorch_release_2.11.0` base =
+torch 2.11.0+rocm7.14/triton 3.7, `--no-build-isolation --no-deps -e .`,
+curated portable deps, `flashlib==0.3.0 --no-deps`, and a `libhiprtc.so`
+dev-symlink the pip-ROCm wheel lacks) serving
+**`orcarouter/Qwen3.8-Flash-Next-Uncensored-FP8`** (~172.8 GiB, 131 shards,
+standard modelopt block-FP8 — 75,264 `weight_scale_inv` tensors, no
+`input_scale`, loaded directly from HF safetensors, no FTW conversion) on
+**ONE R9700 (gfx1201)** — the second card sits idle. Plain `podman run`
+through `scripts/serve-frank-primary.sh` under the root systemd unit
+(`frank-primary.service`, `User=dank`, `LimitMEMLOCK=infinity`); the
+container name is the unit name (`frank-primary`; the vLLM-era
+`frank-primary-qwen` is in the legacy cleanup list). Cutover 2026-10-05;
+rollback tag `frank-primary-vllm-shapeG-20261003` (the tcclaviger vLLM trio
+and its 362-line env lore live in that tag; the tcclaviger image stays in
+dank's rootless storage).
 
-**Unit and self-healing.** Root systemd unit (`scripts/frank-primary.service`,
-`User=dank`, `LimitMEMLOCK=infinity`) — the PLE pinned residency alone is
-54.4 GB, and a root unit grants the pin budget via PID1 (a rootless
-quadlet would be capped by the user manager's limit).
-The serve script (`scripts/serve-frank-primary.sh`) gates the setup state
-(ready + paths), the pinned image (RepoDigests), the PLE byte count
-(`stat -c %s`), every GGUF shard + the mmproj and the GPU device nodes,
-clears a stale container, then execs the fork's `scripts/launch.sh` through
-the shim and **blocks on `docker wait`** — the launcher `--detach`es, so
-without the wait a crashed engine is a green unit; `docker wait` cannot stop
-the container child, hence `ExecStop=podman stop -t 120
-frank-primary-r9v`. The deployer's `deploy_llm_primary` gates state/image/
-PLE/shards, `podman rm -f`s the container, waits out :7770, relinks/enables/
-restarts the unit and health-gates with the MainPID check (the false-green
-stray law).
+**Shape**: all expert weights in **pinned host RAM** (~120 GiB) behind a
+GPU LRU (`--moe-cache-rate 0.10`, ~12 GiB — the single-card ceiling given
+the KV contract), the ~47.7 GiB PLE ngram table pinned in host RAM
+(→ root unit + memlock law), **KV pool `--num-tokens 299968`**
+(24 KiB/token f16 across the 12 full-attention layers ≈ 7.0 GiB; **must be
+a multiple of 64** — the auto attention backend is `qsa_sparse`, which
+overrides page size to 64; 300,000 was refused at first boot), 2 decode
+slots, `--max-output-tokens 36864` (the harness maxTokens; requests that
+omit one get this default). **Parsers are AUTO**: thinking lands in
+`reasoning_content` natively, tool calls in the qwen3.8 dialect — models.yml
+flips `reasoningContentField` to `reasoning_content` on the harness side.
+Vision works (models.yml keeps `input: [text, image]`); the vision tower's
+blocks stream from host RAM two at a time by default.
 
-Verified at cutover (2026-09-27): q8 setup state `ready: true`, all 20
-payload artifacts sha256-verified (175.30 GiB); the scratch-port bring-up
-battery answered vision ("Red" for a 64×64 red PNG), MTP4 mean acceptance
-length 3.19–5.00, and the PLE pinned proof (`mapping_bytes =
-mapping_rss_bytes = 54,400,261,120`, fully RSS-resident); wire + tool
-round-trip + template probe re-verified on :7770 after the deploy. No
-throughput bench was run (Frank declined); the probe-era 11 tok/s decode
-was residency-bound (halved expert headroom at double the bytes/expert),
-not the production number.
+**Measured at cutover**: boot ~2.3 min cold-ish / 92 s warm → /health
+status ok; decode **17.1 tok/s** single-stream temp 0 (the vLLM incumbent
+measured ~30 tok/s on TWO cards; the KV contract caps the single-card LRU,
+and **TP=2 exists but does not boot on ROCm** — the TP allreduce JIT
+(`pynccl.cu`) can't resolve `cuda_runtime.h` for gfx1201; probed
+2026-10-06, signature in the runbook). `/metrics` is a 404 — the Dankclaw
+Prometheus llm-fleet scrape of :7770 reads down until repointed.
+`ft ctl cache --kv/--moe` resizes pools live (grow only; shrink needs a
+restart); `ft ctl stats` reports occupancy.
 
-Runbook: `serverAppConfigs-docs/LLM/frank-primary-r9v-radiance-qwen38-q8.md`.
-Rollback: the uncensored payload — flip `R9V_STATE_FILE` back to
-`/home/dank/r9v-state/uncensored/setup.json`, drop the two q8-only keys
-and the template keys, redeploy (state + `restore-config-20260927.env` +
-payload intact on disk) — see
-[[frank-primary-r9v-radiance-qwen38-uncensored]].
+**Deploy gate law (changed for FreeToken)**: `/health` returns **200 with
+`{"status":"loading"}`** while the engine loads — the FastAPI front end
+binds before the scheduler is up, so a bare `curl` false-greens (measured:
+"healthy after 6 s" mid-page-in). The deployer loops until the body says
+`"status":"ok"`; the MainPID stray-port guard is unchanged. The image gate
+is `podman image exists` (a local build carries no RepoDigests — the old
+digest grep refused exactly this shape; pinning lives in the Dockerfile).
+
+Runbook: `serverAppConfigs-docs/LLM/frank-primary-freetoken-pr587.md`.
+Retired vLLM era: `serverAppConfigs-docs/LLM/dankserver-qwen38-fp8-expert-offload-tuning.md`.
+Retired GLM experiment: `serverAppConfigs-docs/LLM/frank-primary-glm53-flash-w4a16.md`.
 
 ## Key Lessons Learned
 Durable operational lessons from the fleet's history. Full detail in the
@@ -590,15 +671,56 @@ linked runbooks.
   the KV pool is worse than the OOM it prevents.
 - Host swap state is invisible to `deploy.sh` (no repo file) — check
   `swapon --show` before reasoning about headroom.
+- **Silent RAM corruption outranks every software theory.** frank-primary's
+  ~40k-token output corruption (2026-10-04/05, repetition/garbage in long
+  agent conversations) was marginal DDR5 at 4800 MT/s on four DIMMs, fixed
+  by 4600. Consumer non-ECC memory raises **no MCE** for a silent flip, and
+  a short in-band pattern check (60 s, 8 GiB of 246 GiB) is not a stability
+  proof — combined, those two instruments said "ruled out" while the RAM
+  was the cause. "4800 is the JEDEC floor on 5600-rated DIMMs" is a rating,
+  not a margin guarantee. When the fault is data-shaped (NaN states,
+  garbage tokens) and a software revert doesn't move it, price in the RAM
+  before the runtime — and remember a revert tested under the same marginal
+  hardware is itself confounded. See the runbook's 2026-10-05 section.
 
 **Runtime-specific guards**
 
+- **Rootless podman cannot RAISE rlimits, and `--ulimit nofile=1048576`
+  dies at container create, not at runtime.** runc reports
+  `error setting rlimit type 7: operation not permitted` (type 7 =
+  RLIMIT_NOFILE) when the requested hard limit exceeds the host process's
+  hard limit — on Dankclaw that ceiling is 524288. Inherited-equal limits
+  are fine, so `memlock=-1:-1` works iff the unit grants
+  `LimitMEMLOCK=infinity` (systemd is privileged; the value flows through
+  podman to runc without any raise). Upstream halogen's own run lines pass
+  NO `--ulimit` at all — copy vendor run lines before adding flags
+  (measured 2026-10-05 on the frank-large cutover).
 - **Kairic Edge** (if ever re-deployed): the pinned CK commit needs
   `third_party/composable_kernel/include/ck/config.h` generated manually
   (untracked; `git clean` removes it), and its greedy-argmax fastpath rejects
   non-greedy requests and reasoning budgets. Re-test chat before production.
 - **FastMTP sidecar**: build must be commit `4df29be4` + the HauhauCS patch;
   stock binaries fail the draft load with a tensor-shape mismatch.
+- **A GGUF's custom tensor type picks the runtime — read the histogram before
+  promising a file to a loader.** `Qwen3.8-27B-Q8_0_ROCMFPX_AGENT.gguf` is
+  ftype 115 with 311 tensors of type 103 (`GGML_TYPE_Q8_0_ROCMFPX`); stock
+  llama.cpp and gufo refuse it at load, so ROCmFPX was not a preference but
+  the only door. Counting tensor types out of the GGUF KV section takes one
+  python read and settles a runtime argument in minutes. The build identity
+  matters as much as the fork: the 2026-08-25 ROCmFPX builds SIGSEGV in
+  `ggml_compute_forward_set_rows` on this same file, so
+  `scripts/build-rocmfpx.sh` pins commit `36e9acd` by SHA.
+- **Speculative-decode flags move as a SET.** Handing the ROCmFPX fork
+  `--model-draft` with no `--spec-type` (the naive way to build a "no MTP"
+  bench arm) segfaulted at load, core dumped, no diagnostic (2026-10-02). A
+  baseline arm must strip `--model-draft` AND every `--spec-draft-*` AND
+  `--spec-type`, or it measures a crash instead of the absence of drafting.
+- **An orphaned scratch server taxes the whole host.** A bench child left
+  alive by a client-side exception kept ~26 GiB of weights resident on a
+  121 GiB box, pushed `available` to 3 GiB, and every later rate was suspect.
+  Record the child pid and kill it in a `finally`, then prove `free` came back
+  before believing a number. Kill by pid — never `pkill -f llama-server`,
+  which also hits frank-small.
 
 **llama.cpp serving**
 
@@ -608,19 +730,26 @@ linked runbooks.
   the MAIN model's weights (`shares_model`), so there is no second copy of a
   12 GiB model in VRAM and no `-md` argument. The plain build has no MTP
   tensors at all, and pointing `--model-draft` at the file would load it twice.
-- **What can be chained, and what cannot.** `--spec-type a,b` builds every
-  enabled implementation and chains drafting through them, in a fixed internal
-  order (ngram types first, then the drafters — CLI order is irrelevant). The
-  gate is `ctx_dft`: **draftless** types (the ngram family) are added
-  unconditionally, while `draft-mtp`, `draft-dflash`, `draft-dspark` and
-  `draft-eagle3` all require a draft CONTEXT. So a chain may hold one
-  context-backed drafter plus any number of ngram types — `draft-dflash,ngram-mod`
-  is legal and measured (2.6x the drafter alone on prompt-echo traffic, byte-
-  identical to it on prose at the fleet's 48/64 window) — but
-  `draft-mtp,draft-dflash` dies at load: *"context type MTP requested but model
-  doesn't contain MTP layers"*, because MTP builds its context against whatever
-  model the draft context holds. Two context-backed types cannot share one
-  process, and MTP's head only loads when MTP **is** the draft model.
+- **What can be chained, what cannot, and what a chain costs on a hybrid.**
+  `--spec-type a,b` builds every enabled implementation and chains drafting
+  through them, in a fixed internal order (ngram types first, then the
+  drafters — CLI order is irrelevant). The gate is `ctx_dft`: **draftless**
+  types (the ngram family) are added unconditionally, while `draft-mtp`,
+  `draft-dflash`, `draft-dspark` and `draft-eagle3` all require a draft
+  CONTEXT. So a chain may hold one context-backed drafter plus any number of
+  ngram types — `draft-dflash,ngram-mod` is legal and measured (2.6x the
+  drafter alone on prompt-echo traffic, byte-identical to it on prose at the
+  fleet's 48/64 window) — but `draft-mtp,draft-dflash` dies at load: *"context
+  type MTP requested but model doesn't contain MTP layers"*, because MTP
+  builds its context against whatever model the draft context holds. Two
+  context-backed types cannot share one process, and MTP's head only loads
+  when MTP **is** the draft model. **A LEGAL chain can still lose on a
+  recurrent model:** `draft-mtp,ngram-mod` on frank-large's qwen35 hybrid
+  measured −15 % prose / −7 % block-copy (2026-10-03) — adding the ngram layer
+  logs `recurrent state rollback is not compatible with 'ngram-mod' -
+  disabling rollback support`, and MTP verification depends on that rollback,
+  so acceptance drops on ALL traffic even where the matcher fires. Measure
+  chains per architecture, never by analogy.
 - **A drafter's KV scales with the TARGET's context, and its window policy
   decides how much that costs.** The draft context is built with
   `cparams.n_ctx = llama_n_ctx(ctx_tgt)` (`common/speculative.cpp`) — there is
@@ -637,7 +766,8 @@ linked runbooks.
   trip moved the balance point FOUR layers (~4.3 GiB of headroom, 40:25 ->
   36:29), which is the practical size of the effect. See
   [[frank-medium-llamacpp-dflash2]].
-- **`--ubatch-size` has an OPTIMUM on this stack, not a plateau — and its
+- **`--ubatch-size` has an OPTIMUM on the 2x R9700 layer split, not a plateau —
+  and the curve is BACKEND-SPECIFIC, so never carry it across (next bullet). Its
   compute buffers land about TWICE as heavy on the card that holds the output
   layers and the drafter.** Measured on frank-medium 2026-09-16 at the deployed
   shape (262144 over two slots, a 64k prefill measured at 64k depth,
@@ -655,6 +785,19 @@ linked runbooks.
   its 2020 MiB pp buffer here under the DSpark drafter, when ROCm1 had 0.08 GiB
   free. That was a CONTIGUITY failure, not a total-bytes one, and no totals
   model — the sweep's or the seam's — can see it.)
+- **On the CUDA 3090 the same knob has NO interior optimum up to 2048, so
+  frank-small ships 4096/2048.** Measured 2026-10-05 by boot-pair A/B on the
+  live service: at 8,928 prompt tokens ubatch 1024 gives 3,972 / 4,077 t/s and
+  ubatch 2048 gives 4,207 / 4,212 — **+4.6 %** — for **616 MiB** (free 2,849 →
+  2,279). Decode did not move (91.9–93.0 vs 93.8–94.3, i.e. noise), and a
+  279-token prompt, which is below either ubatch, came back identical — that is
+  the mechanical proof the knob was the only variable, since a prompt shorter
+  than the ubatch cannot see it. The logical `--batch-size` was again free. And
+  the predecessor's "ubatch 2048 failed the scratch boot" was a shape
+  observation, not a law: the same file boots 2048 clean at 2 × 200,704. Bench
+  it by restarting the SERVICE through a transient `FRANK_SMALL_ENV_FILE`
+  drop-in, never a scratch port — one card holds one server. See
+  [[frank-small-llama-mimo-q6k-2x200704]].
 - **One `-ts` step is one LAYER, and a hybrid stack's layers are not the same
   size.** qwen35 is 16 x (3 Gated DeltaNet + 1 Gated Attention), so a step is
   worth ~0.43 GiB on a DeltaNet boundary and ~1.43 GiB on a full-attention one
@@ -671,6 +814,41 @@ linked runbooks.
   host-RAM hash table) and its window decides which traffic it can see: too
   short and it proposes on ordinary prose fragments, which costs verification
   without acceptance.
+- **`--parallel` divides the pool, so MORE conversations cost WINDOW, and the
+  per-slot number is the contract — not `ctx-size`.** `n_ctx_slot =
+  CTX_SIZE / PARALLEL`, and llama.cpp refuses a prompt longer than the SLOT
+  (`400 exceed_context_size_error: request (184001 tokens) exceeds the
+  available context size (180224 tokens)`) even when the pool has room. So
+  frank-small's 2 → 3 reslice moved each conversation from the model's native
+  262,144 down to 180,224, and `models.yml` `contextWindow` had to follow or
+  the harness would discover the ceiling as a 400 on a long conversation. Side
+  effect worth knowing: the shorter slot made decode ~5 % FASTER (106.8 vs
+  100.4 tok/s) — less attention to scan and one fewer DeltaNet state resident.
+  The 2026-10-03 reslice went the other way (3 → 2 slots, 180,224 → 200,704,
+  pool 401,408) to fund unquantized f16 V, and `contextWindow` followed again;
+  decode landed 104.8 tok/s — the ~2 % off 106.8 is the exact-V tax. The same
+  day the freed space also bought the Q6_K rung (now 94.4 tok/s — the rung
+  itself is a further measured −10 %, weight-bandwidth, not slot geometry).
+  All moves measured; both slot directions cost or buy window, never pool.
+  See [[frank-small-llama-mimo-q6k-2x200704]].
+- **Price a VRAM budget from two boots, never from a formula — and an mmproj
+  costs ~1.5x its file size.** The frank-small 3090 fit was solved from two
+  measurements: 2 × 262,144 with no projector = 20,190 MiB used (KV 12,800 ⇒
+  **non-KV 7,390**), 3 × 196,608 with the f16 projector = 22,972 used (KV
+  14,113 ⇒ **non-KV 8,859**). The difference is the vision tower: a 918 MB file
+  that bills ~1.4 GiB, because the encoder's activation and compute buffers
+  ride with it. Subtracting the two rows is exact; extrapolating weights+KV+
+  "a bit for buffers" is what leaves 1.1 GiB free when you meant 2.7. Then
+  solve `KV ≤ card − non-KV − headroom` and round to a multiple of
+  `PARALLEL × 4096`. Recorded in `scripts/frank-small.env` so the next resize
+  starts from measurement.
+- **`images` and `image_url` are NOT interchangeable: only the content part
+  reaches the model.** `--mmproj` loads, `usage.prompt_tokens` rises, and the
+  model still invents an answer if the client used the legacy top-level
+  `images` field on `/v1/chat/completions` — frank-small answered "Brown." for
+  a solid-red PNG that way and "Red." through the OpenAI `image_url` part. The
+  failure is silent because a hallucinated color is a plausible answer. Probe
+  vision with a solid color and expect an exact word.
 - **`--jinja` is what `tools` needs.** On 2026-09-16 a subagent died with
   `500 tools param requires --jinja flag`, from the ik_llama.cpp runtime that
   then served frank-large: every tool-using agent on that endpoint was broken
@@ -771,9 +949,9 @@ Active:
 
 | Service | Runbook |
 |---------|---------|
-| frank-large (stock llama.cpp Qwen3.8-27B UD-Q8_K_XL, Dankclaw) | `serverAppConfigs-docs/LLM/frank-large-unsloth-qwen38-27b-q8xl.md` |
-| frank-small (llama.cpp Gemma4-12B QAT, Dankclaw) | `serverAppConfigs-docs/LLM/frank-small-gemma4-12b.md` |
-| frank-primary (R9V radiance-vLLM Qwen3.8-Flash-Next Q8_0, Dankserver) | `serverAppConfigs-docs/LLM/frank-primary-r9v-radiance-qwen38-q8.md` |
+| frank-large (halogen container, Qwen3.8-27B .hgn + DFlash2, 1×200,700, text-only, Dankclaw) | `serverAppConfigs-docs/LLM/frank-large-halogen27b-dflash2.md` |
+| frank-small (llama.cpp CUDA, MiMo-V2.6-Distill-Qwen-9B Q6_K, 2×200,704 f16 KV, Dankclaw) | `serverAppConfigs-docs/LLM/frank-small-llama-mimo-q6k-2x200704.md` |
+| frank-primary (FreeToken PR #587 ROCm, Qwen3.8-Flash-Next FP8, 1×R9700, Dankserver) | `serverAppConfigs-docs/LLM/frank-primary-freetoken-pr587.md` |
 | Dankclaw kernel posture (`amd_iommu=off`) | `serverAppConfigs-docs/LLM/dankclaw-iommu-translation-off.md` |
 | Deployment flow + incidents | `serverAppConfigs-docs/LLM/deployment.md` |
 
@@ -781,6 +959,16 @@ Retired (rollback targets and history):
 
 | Runtime | Host | Window | Runbook |
 |---------|------|--------|---------|
+| ROCmFPX llama.cpp fork Qwen3.8-27B Uncensored Q6_0_ROCMFPX_AGENT + MTP (`frank-large`) | Dankclaw | 2026-10-03 → 10-05 (rollback target, payload + fork build on disk) | `frank-large-rocmfpx-q6agent-uncensored-mtp.md` |
+| tcclaviger vLLM fork Qwen3.8-Flash-Next-Uncensored FP8, TP=2, shape G (`frank-primary`) | Dankserver | 2026-09-29 → 10-05 (rollback target — tag `frank-primary-vllm-shapeG-20261003`; trio + env lore in the tag, image in rootless storage) | `dankserver-qwen38-fp8-expert-offload-tuning.md` |
+| ROCmFPX llama.cpp fork Qwen3.8-27B Q8_0_ROCMFPX_AGENT + MTP (`frank-large`) | Dankclaw | 2026-10-02 → 10-03 (rollback target, payload on disk) | `frank-large-rocmfpx-q8agent-mtp.md` |
+| gufo v0.1.1 Qwen3.8-27B UD-Q8_K_XL/UD-Q4_K_XL + DFlash2 (`frank-large`) | Dankclaw | 2026-09-28 → 10-01, again 10-01 → 10-02 (payloads removed from disk 2026-10-03; rollback = re-download) | `frank-large-gufo-qwen38-27b.md` |
+| vLLM TheRock MiMo-V2.6-Distill-Qwen-9B BF16 + cross-model MTP (`frank-large`) | Dankclaw | 2026-10-01 (same-day rollback to gufo) | `frank-large-vllm-mimo-v26-bf16.md` |
+| llama.cpp master MiMo-V2.6-Distill-Qwen-9B Q6_K (`frank-small`) | Dankclaw | 2026-10-01 → 10-02 (the rung returned ACTIVE 2026-10-03 at the 2×200,704 shape — `frank-small-llama-mimo-q6k-2x200704.md`) | `frank-small-llama-mimo-q6k.md` |
+| llama.cpp master MiMo-V2.6-Distill-Qwen-9B Q5_K_M 2×200,704 f16 KV (`frank-small`) | Dankclaw | 2026-10-02 → 10-03 (rollback target — one `MODEL_FILE` line, buys back 875 MiB + the −10 % rung tax; owns the boot-pair table) | `frank-small-llama-mimo-q5km.md` |
+| SGLang MiMo-V2.6 INT8-W8A16-AutoRound (`frank-small`) | Dankclaw | 2026-10-01 (same-day swap to the GGUF path) | `frank-small-sglang-mimo-v26.md` |
+|tcclaviger vLLM GLM-5.3-Flash W4A16 int4, +/- DFlash2 drafter (`frank-primary`)|Dankserver|2026-09-30 → 10-01|`frank-primary-glm53-flash-w4a16.md`|
+| llama.cpp b11030 Qwen3.8-27B UD-Q8_K_XL + DFlash2 (`frank-large`) | Dankclaw | 2026-09-21 → 09-28 | `frank-large-unsloth-qwen38-27b-q8xl.md` |
 | R9V radiance-vLLM Qwen3.8-Flash-Next-Uncensored IQ4_XS (`frank-primary`) | Dankserver | 2026-09-27 (same-day swap to Q8_0) | `frank-primary-r9v-radiance-qwen38-uncensored.md` |
 | llama.cpp b11030 Qwen3.8-Flash-Next Q8_0 + shared MTP (`frank-primary`) | Dankserver | 2026-09-26 → 09-27 | `frank-primary-llamacpp-qwen38-flash-next-q8.md` |
 | ik_llama.cpp Qwen3.8-Flash-Next-Uncensored i1-Q4_K_M (`frank-large`) | Dankclaw | 2026-09-15 → 09-16 | `frank-large-ik-uncensored.md` |
@@ -793,6 +981,7 @@ Retired (rollback targets and history):
 | Kairic Edge Qwen3.8-27B IU4 W4A4 (`frank-large`) | Dankclaw | 2026-09-21 (same-day swap) | `frank-large-kairic-edge-qwen38-27b.md` |
 | ISTA-DASLab Qwen3.8-27B-GSQ-RCO IQ3_S (`frank-small`) | Dankclaw | 2026-09-16 → 09-17 | `frank-small-qwen38-gsq-rco.md` |
 | vLLM Gemma4-12B w4a16 (`frank-small`) | Dankclaw | 2026-09-19 (same-day revert) | `frank-small-vllm-gemma4.md` |
+| llama.cpp Gemma4-12B QAT Q4_K_M (`frank-small`) | Dankclaw | 2026-09-17 → 10-01 | `frank-small-gemma4-12b.md` |
 | agentionai ROCmFP4-FAST imatrix | Dankclaw | 2026-08-29 → 08-30 | `agentionai-fast-imatrix.md` |
 | eemin Qwen3.8-27B ROCmI4 W4A4 MTP | Dankclaw | 2026-08-18 → 08-25 | `eemin-rocmi4-mtp.md` |
 | Kairic Edge IU4 | Dankclaw | 2026-08-22 (same-day revert) | `kairic-edge.md` |

README.md

@@ -13,7 +13,7 @@ one deploy script.
 | Host | User | IP | Podman stack | LLM service | LLM port |
 |------|------|----|--------------|-------------|----------|
 | dankserver | dank | 192.168.0.235 | media and library | frank-medium (ROCmFPX llama.cpp, Qwen3.8-27B AGENT) | 7770 |
-| dankclaw | dank | 192.168.0.143 | AI, search, development, observability | frank-large (CIRU ROCmFPX, Qwen3.8-27B) | 7770 |
+| dankclaw | dank | 192.168.0.143 | AI, search, development, observability | frank-large (gufo, Qwen3.8-27B UD-Q8_K_XL) | 7770 |
 
 ## Active Services
 

autoresearch.sh

@@ -0,0 +1,117 @@
+#!/usr/bin/env bash
+# autoresearch.sh — canonical benchmark entrypoint: speed of frank-primary.
+#
+# Runs on the operator box and drives one bench window on DANKSERVER:
+#   1. fingerprint the candidate trio (env + serve script + unit)
+#   2. memo fast-path: an identical fingerprint that already produced a FULLY
+#      successful window (bench + verified production restore) returns the
+#      cached metrics without touching the fleet
+#   3. stage the candidate trio plus the measure client into
+#      /home/dank/.autoresearch-frank-primary/ (ephemeral; the repo checkout
+#      on the host stays the production source of truth)
+#   4. run the window (scripts/autoresearch/frank-primary-window.sh):
+#      stop + mask the production unit, boot the staged config AS THE SERVICE
+#      (same serve script, same container, same port 7770), wait for health,
+#      run the in-container usage-counter measure, then restore the production
+#      unit and verify its health — AGENTS.md Benchmarking Policy
+#   5. print METRIC lines (parse by run_experiment)
+#
+# Primary metric: decode_tps (HIGHER is better) — median across the three
+# prompt depths (115/1056/4096) of per-depth medians of
+# completion_tokens / (wall - ttft), 3 runs each after a warm pass,
+# temperature 0, the campaign's deterministic word-seed workload.
+#
+# Exit 0 with METRIC lines on success; non-zero on any failure (the window
+# script's EXIT trap restores production on every path).
+set -euo pipefail
+
+SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
+HOST="${FRANK_PRIMARY_AR_HOST:-dankserver}"
+STAGE="/home/dank/.autoresearch-frank-primary"
+
+# FRANK_PRIMARY_AR_ENV_FILE stages a DIFFERENT env than the repo's — that is how
+# the arms driver A/Bs engine knobs without mutating the checkout. The
+# fingerprint covers whatever file is staged, so the memo cannot cross arms.
+ENV_FILE="${FRANK_PRIMARY_AR_ENV_FILE:-$SCRIPT_DIR/scripts/frank-primary.env}"
+SERVE="$SCRIPT_DIR/scripts/serve-frank-primary.sh"
+UNIT="$SCRIPT_DIR/scripts/frank-primary.service"
+MEASURE="$SCRIPT_DIR/scripts/autoresearch/frank-primary-measure.py"
+WINDOW="$SCRIPT_DIR/scripts/autoresearch/frank-primary-window.sh"
+
+for f in "$ENV_FILE" "$SERVE" "$UNIT" "$MEASURE" "$WINDOW"; do
+    [ -f "$f" ] || { echo "[autoresearch] missing harness file: $f" >&2; exit 1; }
+done
+command -v python3 >/dev/null 2>&1 || { echo "[autoresearch] python3 required on operator box" >&2; exit 1; }
+
+# Fingerprint of everything that can change the engine's argv or environment.
+HASH="$(cat "$ENV_FILE" "$SERVE" "$UNIT" | sha256sum | cut -d' ' -f1)"
+echo "[autoresearch] candidate fingerprint $HASH" >&2
+
+ssh_run() { ssh -o BatchMode=yes -o ConnectTimeout=15 "$HOST" "$@"; }
+
+emit_metrics() { # $1 = run.json
+    python3 -c '
+import json, sys
+run = json.loads(sys.argv[1])
+m = run["measure"]
+print("METRIC decode_tps=%s" % m["decode_tps"])
+for k in ("decode_tps_115", "decode_tps_1056", "decode_tps_4096"):
+    print("METRIC %s=%s" % (k, m[k]))
+print("METRIC ttft_ms_4096=%s" % m["ttft_ms_4096"])
+print("METRIC kv_tokens=%s" % (run.get("kv_tokens") or 0))
+print("METRIC boot_seconds=%s" % (run.get("boot_seconds") or -1))
+print("METRIC completion_tokens_4096=%s" % m.get("completion_tokens_4096", 0))
+# Any extra sampling temperature the measure swept (AR_TEMPS), e.g. the 0.7
+# arm — that is where speculative-decode acceptance shows up, since a greedy
+# draft matches a greedy target exactly and is rejected more under real
+# sampling. Suffixed keys only (_t0.7), so the historical lines are untouched.
+for k in sorted(m):
+    if "_t" in k and (k.startswith("decode_tps") or k.startswith("ttft_ms")):
+        print("METRIC %s=%s" % (k, m[k]))
+' "$1"
+}
+
+# --- memo fast-path -------------------------------------------------------
+cached_hash="$(ssh_run "cat '$STAGE/last/hash' 2>/dev/null" 2>/dev/null || true)"
+if [ -n "$cached_hash" ] && [ "$cached_hash" = "$HASH" ]; then
+    cached_run="$(ssh_run "cat '$STAGE/last/run.json' 2>/dev/null" || true)"
+    if [ -n "$cached_run" ]; then
+        echo "[autoresearch] memo hit — cached metrics for $HASH, fleet untouched" >&2
+        emit_metrics "$cached_run"
+        exit 0
+    fi
+fi
+
+# --- stage the candidate ----------------------------------------------------
+echo "[autoresearch] staging candidate on $HOST:$STAGE" >&2
+ssh_run "mkdir -p '$STAGE'"
+scp -q "$ENV_FILE" "$HOST:$STAGE/frank-primary.env"
+scp -q "$SERVE" "$HOST:$STAGE/serve.sh"
+scp -q "$MEASURE" "$HOST:$STAGE/measure.py"
+scp -q "$WINDOW" "$HOST:$STAGE/window.sh"
+ssh_run "printf '%s\n' '$HASH' > '$STAGE/hash'"
+
+# --- the window -------------------------------------------------------------
+# Budgets: bench boot up to 20 min (warm boots measured ~7 min), measure up
+# to 30 min, restore boot up to 20 min — call run_experiment with
+# timeout_seconds >= 3600.
+echo "[autoresearch] opening bench window (stop+mask -> boot -> measure -> restore)" >&2
+set +e
+ssh_run "bash '$STAGE/window.sh'"
+rc=$?
+set -e
+if [ "$rc" -ne 0 ]; then
+    echo "[autoresearch] window failed (rc=$rc); host-side evidence:" >&2
+    ssh_run "tail -60 '$STAGE/fail.log' 2>/dev/null || tail -40 '$STAGE/serve.out' 2>/dev/null || tail -40 '$STAGE/measure.out' 2>/dev/null || true" >&2 || true
+    exit 1
+fi
+
+run_json="$(ssh_run "cat '$STAGE/run.json'")" || {
+    echo "[autoresearch] could not read $STAGE/run.json back" >&2; exit 1; }
+echo "[autoresearch] window result:" >&2
+printf '%s\n' "$run_json" >&2
+emit_metrics "$run_json"
+
+echo "[autoresearch] engine evidence (plan / compile path / spec counters):" >&2
+ssh_run "cat '$STAGE/evidence.log' 2>/dev/null" >&2 || true
+exit 0

echo_prompt.txt

@@ -0,0 +1,10 @@
+Below is a telemetry API schema, labeled SCHEMA A. Later in this message the same schema is repeated word for word, labeled SCHEMA B.
+Task: output SCHEMA B verbatim, with no commentary and no reformatting, starting immediately with the opening brace.
+
+SCHEMA A:
+{"version":"2.3.1","endpoints":[{"id":"ep-000","path":"/v2/telemetry/gw5978x49","method":"GET","latency_ms":219,"owner":"team-0","limit":{"burst":221,"sustained":149}},{"id":"ep-001","path":"/v2/telemetry/gw6279x77","method":"POST","latency_ms":456,"owner":"team-1","limit":{"burst":227,"sustained":173}},{"id":"ep-002","path":"/v2/telemetry/gw4501x82","method":"PUT","latency_ms":488,"owner":"team-2","limit":{"burst":407,"sustained":189}},{"id":"ep-003","path":"/v2/telemetry/gw3780xA8","method":"DELETE","latency_ms":90,"owner":"team-3","limit":{"burst":146,"sustained":132}},{"id":"ep-004","path":"/v2/telemetry/gw6491x17","method":"GET","latency_ms":842,"owner":"team-4","limit":{"burst":488,"sustained":193}},{"id":"ep-005","path":"/v2/telemetry/gw4848xAC","method":"POST","latency_ms":321,"owner":"team-5","limit":{"burst":116,"sustained":51}},{"id":"ep-006","path":"/v2/telemetry/gw3414x06","method":"PUT","latency_ms":51,"owner":"team-6","limit":{"burst":126,"sustained":122}},{"id":"ep-007","path":"/v2/telemetry/gw2193x74","method":"DELETE","latency_ms":428,"owner":"team-0","limit":{"burst":454,"sustained":162}},{"id":"ep-008","path":"/v2/telemetry/gw4185xB7","method":"GET","latency_ms":717,"owner":"team-1","limit":{"burst":197,"sustained":127}},{"id":"ep-009","path":"/v2/telemetry/gw7547x3E","method":"POST","latency_ms":155,"owner":"team-2","limit":{"burst":336,"sustained":177}},{"id":"ep-010","path":"/v2/telemetry/gw1090xE4","method":"PUT","latency_ms":772,"owner":"team-3","limit":{"burst":441,"sustained":198}},{"id":"ep-011","path":"/v2/telemetry/gw2746xC7","method":"DELETE","latency_ms":439,"owner":"team-4","limit":{"burst":113,"sustained":46}},{"id":"ep-012","path":"/v2/telemetry/gw9486x76","method":"GET","latency_ms":55,"owner":"team-5","limit":{"burst":286,"sustained":64}},{"id":"ep-013","path":"/v2/telemetry/gw2988x74","method":"POST","latency_ms":140,"owner":"team-6","limit":{"burst":411,"sustained":119}},{"id":"ep-014","path":"/v2/telemetry/gw9702xFC","method":"PUT","latency_ms":576,"owner":"team-0","limit":{"burst":305,"sustained":82}},{"id":"ep-015","path":"/v2/telemetry/gw8251x9C","method":"DELETE","latency_ms":838,"owner":"team-1","limit":{"burst":369,"sustained":130}},{"id":"ep-016","path":"/v2/telemetry/gw7991xD4","method":"GET","latency_ms":565,"owner":"team-2","limit":{"burst":229,"sustained":41}},{"id":"ep-017","path":"/v2/telemetry/gw8777x73","method":"POST","latency_ms":269,"owner":"team-3","limit":{"burst":385,"sustained":64}},{"id":"ep-018","path":"/v2/telemetry/gw5543xC4","method":"PUT","latency_ms":800,"owner":"team-4","limit":{"burst":267,"sustained":125}},{"id":"ep-019","path":"/v2/telemetry/gw4919x46","method":"DELETE","latency_ms":454,"owner":"team-5","limit":{"burst":40,"sustained":183}},{"id":"ep-020","path":"/v2/telemetry/gw5681x3C","method":"GET","latency_ms":282,"owner":"team-6","limit":{"burst":172,"sustained":82}},{"id":"ep-021","path":"/v2/telemetry/gw9849x14","method":"POST","latency_ms":145,"owner":"team-0","limit":{"burst":78,"sustained":60}},{"id":"ep-022","path":"/v2/telemetry/gw7275xB1","method":"PUT","latency_ms":160,"owner":"team-1","limit":{"burst":362,"sustained":55}},{"id":"ep-023","path":"/v2/telemetry/gw2052x6A","method":"DELETE","latency_ms":421,"owner":"team-2","limit":{"burst":170,"sustained":139}},{"id":"ep-024","path":"/v2/telemetry/gw8633x6A","method":"GET","latency_ms":67,"owner":"team-3","limit":{"burst":106,"sustained":108}},{"id":"ep-025","path":"/v2/telemetry/gw7381xE7","method":"POST","latency_ms":792,"owner":"team-4","limit":{"burst":300,"sustained":179}},{"id":"ep-026","path":"/v2/telemetry/gw1320xDB","method":"PUT","latency_ms":787,"owner":"team-5","limit":{"burst":295,"sustained":98}},{"id":"ep-027","path":"/v2/telemetry/gw8814x01","method":"DELETE","latency_ms":364,"owner":"team-6","limit":{"burst":153,"sustained":193}},{"id":"ep-028","path":"/v2/telemetry/gw7389xDA","method":"GET","latency_ms":859,"owner":"team-0","limit":{"burst":215,"sustained":138}},{"id":"ep-029","path":"/v2/telemetry/gw9947xCC","method":"POST","latency_ms":621,"owner":"team-1","limit":{"burst":460,"sustained":57}},{"id":"ep-030","path":"/v2/telemetry/gw8999x38","method":"PUT","latency_ms":283,"owner":"team-2","limit":{"burst":224,"sustained":125}},{"id":"ep-031","path":"/v2/telemetry/gw1475x63","method":"DELETE","latency_ms":348,"owner":"team-3","limit":{"burst":343,"sustained":174}},{"id":"ep-032","path":"/v2/telemetry/gw7624xB9","method":"GET","latency_ms":173,"owner":"team-4","limit":{"burst":431,"sustained":120}},{"id":"ep-033","path":"/v2/telemetry/gw3091xFB","method":"POST","latency_ms":641,"owner":"team-5","limit":{"burst":274,"sustained":7}},{"id":"ep-034","path":"/v2/telemetry/gw7455x97","method":"PUT","latency_ms":581,"owner":"team-6","limit":{"burst":340,"sustained":7}},{"id":"ep-035","path":"/v2/telemetry/gw2375xA4","method":"DELETE","latency_ms":442,"owner":"team-0","limit":{"burst":70,"sustained":119}},{"id":"ep-036","path":"/v2/telemetry/gw3977x0C","method":Line truncated
+
+SCHEMA B (restated, identical to A):
+{"version":"2.3.1","endpoints":[{"id":"ep-000","path":"/v2/telemetry/gw5978x49","method":"GET","latency_ms":219,"owner":"team-0","limit":{"burst":221,"sustained":149}},{"id":"ep-001","path":"/v2/telemetry/gw6279x77","method":"POST","latency_ms":456,"owner":"team-1","limit":{"burst":227,"sustained":173}},{"id":"ep-002","path":"/v2/telemetry/gw4501x82","method":"PUT","latency_ms":488,"owner":"team-2","limit":{"burst":407,"sustained":189}},{"id":"ep-003","path":"/v2/telemetry/gw3780xA8","method":"DELETE","latency_ms":90,"owner":"team-3","limit":{"burst":146,"sustained":132}},{"id":"ep-004","path":"/v2/telemetry/gw6491x17","method":"GET","latency_ms":842,"owner":"team-4","limit":{"burst":488,"sustained":193}},{"id":"ep-005","path":"/v2/telemetry/gw4848xAC","method":"POST","latency_ms":321,"owner":"team-5","limit":{"burst":116,"sustained":51}},{"id":"ep-006","path":"/v2/telemetry/gw3414x06","method":"PUT","latency_ms":51,"owner":"team-6","limit":{"burst":126,"sustained":122}},{"id":"ep-007","path":"/v2/telemetry/gw2193x74","method":"DELETE","latency_ms":428,"owner":"team-0","limit":{"burst":454,"sustained":162}},{"id":"ep-008","path":"/v2/telemetry/gw4185xB7","method":"GET","latency_ms":717,"owner":"team-1","limit":{"burst":197,"sustained":127}},{"id":"ep-009","path":"/v2/telemetry/gw7547x3E","method":"POST","latency_ms":155,"owner":"team-2","limit":{"burst":336,"sustained":177}},{"id":"ep-010","path":"/v2/telemetry/gw1090xE4","method":"PUT","latency_ms":772,"owner":"team-3","limit":{"burst":441,"sustained":198}},{"id":"ep-011","path":"/v2/telemetry/gw2746xC7","method":"DELETE","latency_ms":439,"owner":"team-4","limit":{"burst":113,"sustained":46}},{"id":"ep-012","path":"/v2/telemetry/gw9486x76","method":"GET","latency_ms":55,"owner":"team-5","limit":{"burst":286,"sustained":64}},{"id":"ep-013","path":"/v2/telemetry/gw2988x74","method":"POST","latency_ms":140,"owner":"team-6","limit":{"burst":411,"sustained":119}},{"id":"ep-014","path":"/v2/telemetry/gw9702xFC","method":"PUT","latency_ms":576,"owner":"team-0","limit":{"burst":305,"sustained":82}},{"id":"ep-015","path":"/v2/telemetry/gw8251x9C","method":"DELETE","latency_ms":838,"owner":"team-1","limit":{"burst":369,"sustained":130}},{"id":"ep-016","path":"/v2/telemetry/gw7991xD4","method":"GET","latency_ms":565,"owner":"team-2","limit":{"burst":229,"sustained":41}},{"id":"ep-017","path":"/v2/telemetry/gw8777x73","method":"POST","latency_ms":269,"owner":"team-3","limit":{"burst":385,"sustained":64}},{"id":"ep-018","path":"/v2/telemetry/gw5543xC4","method":"PUT","latency_ms":800,"owner":"team-4","limit":{"burst":267,"sustained":125}},{"id":"ep-019","path":"/v2/telemetry/gw4919x46","method":"DELETE","latency_ms":454,"owner":"team-5","limit":{"burst":40,"sustained":183}},{"id":"ep-020","path":"/v2/telemetry/gw5681x3C","method":"GET","latency_ms":282,"owner":"team-6","limit":{"burst":172,"sustained":82}},{"id":"ep-021","path":"/v2/telemetry/gw9849x14","method":"POST","latency_ms":145,"owner":"team-0","limit":{"burst":78,"sustained":60}},{"id":"ep-022","path":"/v2/telemetry/gw7275xB1","method":"PUT","latency_ms":160,"owner":"team-1","limit":{"burst":362,"sustained":55}},{"id":"ep-023","path":"/v2/telemetry/gw2052x6A","method":"DELETE","latency_ms":421,"owner":"team-2","limit":{"burst":170,"sustained":139}},{"id":"ep-024","path":"/v2/telemetry/gw8633x6A","method":"GET","latency_ms":67,"owner":"team-3","limit":{"burst":106,"sustained":108}},{"id":"ep-025","path":"/v2/telemetry/gw7381xE7","method":"POST","latency_ms":792,"owner":"team-4","limit":{"burst":300,"sustained":179}},{"id":"ep-026","path":"/v2/telemetry/gw1320xDB","method":"PUT","latency_ms":787,"owner":"team-5","limit":{"burst":295,"sustained":98}},{"id":"ep-027","path":"/v2/telemetry/gw8814x01","method":"DELETE","latency_ms":364,"owner":"team-6","limit":{"burst":153,"sustained":193}},{"id":"ep-028","path":"/v2/telemetry/gw7389xDA","method":"GET","latency_ms":859,"owner":"team-0","limit":{"burst":215,"sustained":138}},{"id":"ep-029","path":"/v2/telemetry/gw9947xCC","method":"POST","latency_ms":621,"owner":"team-1","limit":{"burst":460,"sustained":57}},{"id":"ep-030","path":"/v2/telemetry/gw8999x38","method":"PUT","latency_ms":283,"owner":"team-2","limit":{"burst":224,"sustained":125}},{"id":"ep-031","path":"/v2/telemetry/gw1475x63","method":"DELETE","latency_ms":348,"owner":"team-3","limit":{"burst":343,"sustained":174}},{"id":"ep-032","path":"/v2/telemetry/gw7624xB9","method":"GET","latency_ms":173,"owner":"team-4","limit":{"burst":431,"sustained":120}},{"id":"ep-033","path":"/v2/telemetry/gw3091xFB","method":"POST","latency_ms":641,"owner":"team-5","limit":{"burst":274,"sustained":7}},{"id":"ep-034","path":"/v2/telemetry/gw7455x97","method":"PUT","latency_ms":581,"owner":"team-6","limit":{"burst":340,"sustained":7}},{"id":"ep-035","path":"/v2/telemetry/gw2375xA4","method":"DELETE","latency_ms":442,"owner":"team-0","limit":{"burst":70,"sustained":119}},{"id":"ep-036","path":"/v2/telemetry/gw3977x0C","method":Line truncated
+
+Output SCHEMA B verbatim now:

echo_request.json

@@ -0,0 +1 @@
+{"model": "frank-large", "prompt": "Below is a telemetry API schema, labeled SCHEMA A. Later in this message the same schema is repeated word for word, labeled SCHEMA B.\nTask: output SCHEMA B verbatim, with no commentary and no reformatting, starting immediately with the opening brace.\n\nSCHEMA A:\n{\"version\":\"2.3.1\",\"endpoints\":[{\"id\":\"ep-000\",\"path\":\"/v2/telemetry/gw5978x49\",\"method\":\"GET\",\"latency_ms\":219,\"owner\":\"team-0\",\"limit\":{\"burst\":221,\"sustained\":149}},{\"id\":\"ep-001\",\"path\":\"/v2/telemetry/gw6279x77\",\"method\":\"POST\",\"latency_ms\":456,\"owner\":\"team-1\",\"limit\":{\"burst\":227,\"sustained\":173}},{\"id\":\"ep-002\",\"path\":\"/v2/telemetry/gw4501x82\",\"method\":\"PUT\",\"latency_ms\":488,\"owner\":\"team-2\",\"limit\":{\"burst\":407,\"sustained\":189}},{\"id\":\"ep-003\",\"path\":\"/v2/telemetry/gw3780xA8\",\"method\":\"DELETE\",\"latency_ms\":90,\"owner\":\"team-3\",\"limit\":{\"burst\":146,\"sustained\":132}},{\"id\":\"ep-004\",\"path\":\"/v2/telemetry/gw6491x17\",\"method\":\"GET\",\"latency_ms\":842,\"owner\":\"team-4\",\"limit\":{\"burst\":488,\"sustained\":193}},{\"id\":\"ep-005\",\"path\":\"/v2/telemetry/gw4848xAC\",\"method\":\"POST\",\"latency_ms\":321,\"owner\":\"team-5\",\"limit\":{\"burst\":116,\"sustained\":51}},{\"id\":\"ep-006\",\"path\":\"/v2/telemetry/gw3414x06\",\"method\":\"PUT\",\"latency_ms\":51,\"owner\":\"team-6\",\"limit\":{\"burst\":126,\"sustained\":122}},{\"id\":\"ep-007\",\"path\":\"/v2/telemetry/gw2193x74\",\"method\":\"DELETE\",\"latency_ms\":428,\"owner\":\"team-0\",\"limit\":{\"burst\":454,\"sustained\":162}},{\"id\":\"ep-008\",\"path\":\"/v2/telemetry/gw4185xB7\",\"method\":\"GET\",\"latency_ms\":717,\"owner\":\"team-1\",\"limit\":{\"burst\":197,\"sustained\":127}},{\"id\":\"ep-009\",\"path\":\"/v2/telemetry/gw7547x3E\",\"method\":\"POST\",\"latency_ms\":155,\"owner\":\"team-2\",\"limit\":{\"burst\":336,\"sustained\":177}},{\"id\":\"ep-010\",\"path\":\"/v2/telemetry/gw1090xE4\",\"method\":\"PUT\",\"latency_ms\":772,\"owner\":\"team-3\",\"limit\":{\"burst\":441,\"sustained\":198}},{\"id\":\"ep-011\",\"path\":\"/v2/telemetry/gw2746xC7\",\"method\":\"DELETE\",\"latency_ms\":439,\"owner\":\"team-4\",\"limit\":{\"burst\":113,\"sustained\":46}},{\"id\":\"ep-012\",\"path\":\"/v2/telemetry/gw9486x76\",\"method\":\"GET\",\"latency_ms\":55,\"owner\":\"team-5\",\"limit\":{\"burst\":286,\"sustained\":64}},{\"id\":\"ep-013\",\"path\":\"/v2/telemetry/gw2988x74\",\"method\":\"POST\",\"latency_ms\":140,\"owner\":\"team-6\",\"limit\":{\"burst\":411,\"sustained\":119}},{\"id\":\"ep-014\",\"path\":\"/v2/telemetry/gw9702xFC\",\"method\":\"PUT\",\"latency_ms\":576,\"owner\":\"team-0\",\"limit\":{\"burst\":305,\"sustained\":82}},{\"id\":\"ep-015\",\"path\":\"/v2/telemetry/gw8251x9C\",\"method\":\"DELETE\",\"latency_ms\":838,\"owner\":\"team-1\",\"limit\":{\"burst\":369,\"sustained\":130}},{\"id\":\"ep-016\",\"path\":\"/v2/telemetry/gw7991xD4\",\"method\":\"GET\",\"latency_ms\":565,\"owner\":\"team-2\",\"limit\":{\"burst\":229,\"sustained\":41}},{\"id\":\"ep-017\",\"path\":\"/v2/telemetry/gw8777x73\",\"method\":\"POST\",\"latency_ms\":269,\"owner\":\"team-3\",\"limit\":{\"burst\":385,\"sustained\":64}},{\"id\":\"ep-018\",\"path\":\"/v2/telemetry/gw5543xC4\",\"method\":\"PUT\",\"latency_ms\":800,\"owner\":\"team-4\",\"limit\":{\"burst\":267,\"sustained\":125}},{\"id\":\"ep-019\",\"path\":\"/v2/telemetry/gw4919x46\",\"method\":\"DELETE\",\"latency_ms\":454,\"owner\":\"team-5\",\"limit\":{\"burst\":40,\"sustained\":183}},{\"id\":\"ep-020\",\"path\":\"/v2/telemetry/gw5681x3C\",\"method\":\"GET\",\"latency_ms\":282,\"owner\":\"team-6\",\"limit\":{\"burst\":172,\"sustained\":82}},{\"id\":\"ep-021\",\"path\":\"/v2/telemetry/gw9849x14\",\"method\":\"POST\",\"latency_ms\":145,\"owner\":\"team-0\",\"limit\":{\"burst\":78,\"sustained\":60}},{\"id\":\"ep-022\",\"path\":\"/v2/telemetry/gw7275xB1\",\"method\":\"PUT\",\"latency_ms\":160,\"owner\":\"team-1\",\"limit\":{\"burst\":362,\"sustained\":55}},{\"id\":\"ep-023\",\"path\":\"/v2/telemetry/gw2052x6A\",\"method\":\"DELETE\",\"latency_ms\":421,\"owner\":\"team-2\",\"limit\":{\"burst\":170,\"sustained\":139}},{\"id\":\"ep-024\",\"path\":\"/v2/telemetry/gw8633x6A\",\"method\":\"GET\",\"latency_ms\":67,\"owner\":\"team-3\",\"limit\":{\"burst\":106,\"sustained\":108}},{\"id\":\"ep-025\",\"path\":\"/v2/telemetry/gw7381xE7\",\"method\":\"POST\",\"latency_ms\":792,\"owner\":\"team-4\",\"limit\":{\"burst\":300,\"sustained\":179}},{\"id\":\"ep-026\",\"path\":\"/v2/telemetry/gw1320xDB\",\"method\":\"PUT\",\"latency_ms\":787,\"owner\":\"team-5\",\"limit\":{\"burst\":295,\"sustained\":98}},{\"id\":\"ep-027\",\"path\":\"/v2/telemetry/gw8814x01\",\"method\":\"DELETE\",\"latency_ms\":364,\"owner\":\"team-6\",\"limit\":{\"burst\":153,\"sustained\":193}},{\"id\":\"ep-028\",\"path\":\"/v2/telemetry/gw7389xDA\",\"method\":\"GET\",\"latency_ms\":859,\"owner\":\"team-0\",\"limit\":{\"burst\":215,\"sustained\":138}},{\"id\"Line truncated

old/download-frank-large-dflash.sh

@@ -0,0 +1,76 @@
+#!/usr/bin/env bash
+# Download the z-lab Qwen3.5-9B-DFlash drafter for frank-large's DFlash
+# speculative-decode arm (vLLM --speculative-config method=dflash). Public
+# repo — no auth header on the command line.
+#
+# The drafter is a 6-layer qwen3 draft whose dflash_config pins
+# target_layer_ids [1,5,9,13,17,21,25,29] and block_size 16; its hidden_size
+# 4096 / vocab_size 248320 match the MiMo-V2.6-Distill-Qwen-9B target exactly
+# (both preserve the Qwen3.5-9B geometry), which is what lets vLLM load it as
+# the draft for that target.
+#
+# Usage:
+#   scripts/download-frank-large-dflash.sh [DEST_DIR] [--verify-only]
+set -euo pipefail
+
+REPO="z-lab/Qwen3.5-9B-DFlash"
+BASE="https://huggingface.co/${REPO}/resolve/main"
+
+DEST=""
+VERIFY_ONLY=0
+for arg in "$@"; do
+    case "$arg" in
+        --verify-only) VERIFY_ONLY=1 ;;
+        *) DEST="$arg" ;;
+    esac
+done
+DEST="${DEST:-/home/dank/models/Qwen3.5-9B-DFlash}"
+
+# name bytes sha256(- = non-LFS, size-only check)
+FILES=(
+  "model.safetensors 2583816465 0a42274b32554f48de1faa0d42824e9c2ceda649c30ae0a731cddf410dd698c7"
+  "config.json 1222 -"
+)
+
+mkdir -p "$DEST"
+status=0
+for entry in "${FILES[@]}"; do
+  read -r name size sha <<<"$entry"
+  f="$DEST/$name"
+  have=0
+  if [[ -f "$f" ]] && [[ "$(stat -c%s "$f")" == "$size" ]]; then
+    if [[ "$sha" == "-" ]] || sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+      have=1
+    fi
+  fi
+  if (( have )); then
+    echo "OK        $name"
+    continue
+  fi
+  if (( VERIFY_ONLY )); then
+    echo "MISSING   $name"
+    status=1
+    continue
+  fi
+  echo "FETCHING  $name"
+  tmp="$f.part"
+  curl -fsSL --fail-with-body -o "$tmp" "$BASE/$name"
+  mv "$tmp" "$f"
+  if [[ "$(stat -c%s "$f")" != "$size" ]]; then
+    echo "SIZE FAIL $name ($(stat -c%s "$f") != $size)"
+    status=1
+    continue
+  fi
+  if [[ "$sha" != "-" ]] && ! sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+    echo "SHA FAIL  $name"
+    status=1
+    continue
+  fi
+  echo "VERIFIED  $name"
+done
+if (( status == 0 )); then
+  echo "DOWNLOAD_DONE $DEST"
+else
+  echo "DOWNLOAD_FAILED $DEST" >&2
+fi
+exit "$status"

old/download-frank-large-mimo.sh

@@ -0,0 +1,90 @@
+#!/usr/bin/env bash
+# Download the BF16 MiMo-V2.6-Distill-Qwen-9B payload for frank-large
+# (vLLM/kyuz0 gfx1151, Strix Halo iGPU). Public repo — no auth header on the
+# command line (a Bearer token in argv is world-readable via /proc).
+#
+# Per-file FRESH downloads, no resume: the house rule (see
+# download-frank-small.sh); -f/--fail-with-body plus size+sha256 against the
+# Hub LFS oid is what makes an access error fail instead of "succeeding".
+#
+# Usage:
+#   scripts/download-frank-large-mimo.sh [DEST_DIR] [--verify-only]
+#
+# The verified set below is exactly what the 2026-10-01 bench window served:
+# 17.53 GiB of BF16 weights (4 shards, 18,819,720,848 B total with the rest).
+set -euo pipefail
+
+REPO="XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B"
+BASE="https://huggingface.co/${REPO}/resolve/main"
+
+DEST=""
+VERIFY_ONLY=0
+for arg in "$@"; do
+    case "$arg" in
+        --verify-only) VERIFY_ONLY=1 ;;
+        *) DEST="$arg" ;;
+    esac
+done
+DEST="${DEST:-/home/dank/models/MiMo-V2.6-Distill-Qwen-9B-BF16}"
+
+# name bytes sha256(- = non-LFS, size-only check)
+FILES=(
+  "model-00001-of-00004.safetensors 5276436216 aab052180118aee34abc3029b54eaa49096aac606b97d703866b420dceb703c3"
+  "model-00002-of-00004.safetensors 5033171280 7a0486565f06d25ac4628e9dba470dc3f604353471d240d5a0bf7128f64df396"
+  "model-00003-of-00004.safetensors 5234499504 6c73207563d1879bfd6c143a028cc70be68edff56280458c71a43df4240300f4"
+  "model-00004-of-00004.safetensors 3275613848 1379a7cf8c0b8555a39ab65a47e830e0eb45e776b46045734da3afd73c09eea2"
+  "model.safetensors.index.json 69217 -"
+  "config.json 2784 -"
+  "generation_config.json 126 -"
+  "chat_template.jinja 3916 -"
+  "tokenizer.json 19989325 06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523"
+  "tokenizer_config.json 1165 -"
+  "vocab.json 6722759 -"
+  "merges.txt 3353259 -"
+  "preprocessor_config.json 443 -"
+  "processor_config.json 1191 -"
+  "video_preprocessor_config.json 385 -"
+)
+
+mkdir -p "$DEST"
+status=0
+for entry in "${FILES[@]}"; do
+  read -r name size sha <<<"$entry"
+  f="$DEST/$name"
+  have=0
+  if [[ -f "$f" ]] && [[ "$(stat -c%s "$f")" == "$size" ]]; then
+    if [[ "$sha" == "-" ]] || sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+      have=1
+    fi
+  fi
+  if (( have )); then
+    echo "OK        $name"
+    continue
+  fi
+  if (( VERIFY_ONLY )); then
+    echo "MISSING   $name"
+    status=1
+    continue
+  fi
+  echo "FETCHING  $name"
+  tmp="$f.part"
+  curl -fsSL --fail-with-body -o "$tmp" "$BASE/$name"
+  mv "$tmp" "$f"
+  if [[ "$(stat -c%s "$f")" != "$size" ]]; then
+    echo "SIZE FAIL $name ($(stat -c%s "$f") != $size)"
+    status=1
+    continue
+  fi
+  if [[ "$sha" != "-" ]] && ! sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+    echo "SHA FAIL  $name"
+    status=1
+    continue
+  fi
+  echo "VERIFIED  $name"
+done
+if (( status == 0 )); then
+  echo "DOWNLOAD_DONE $DEST"
+else
+  echo "DOWNLOAD_FAILED $DEST" >&2
+fi
+exit "$status"

old/download-frank-large-qwen35.sh

@@ -0,0 +1,85 @@
+#!/usr/bin/env bash
+# Download the source Qwen3.5-9B checkpoint — the ONLY place this model
+# family's MTP head (15 mtp.* tensors) exists. frank-large runs it as the
+# vLLM cross-model MTP draft (MiMo-V2.6-Distill-Qwen-9B itself ships zero
+# mtp tensors; that is why SGLang's NEXTN on frank-small draws 0.00
+# acceptance). Public repo — no auth header on the command line.
+#
+# NOTE the odd upstream shard names (model.safetensors-0000N-of-00004...).
+#
+# Usage:
+#   scripts/download-frank-large-qwen35.sh [DEST_DIR] [--verify-only]
+set -euo pipefail
+
+REPO="Qwen/Qwen3.5-9B"
+BASE="https://huggingface.co/${REPO}/resolve/main"
+
+DEST=""
+VERIFY_ONLY=0
+for arg in "$@"; do
+    case "$arg" in
+        --verify-only) VERIFY_ONLY=1 ;;
+        *) DEST="$arg" ;;
+    esac
+done
+DEST="${DEST:-/home/dank/models/Qwen3.5-9B}"
+
+# name bytes sha256(- = non-LFS, size-only check)
+FILES=(
+  "model.safetensors-00001-of-00004.safetensors 5276436216 db6f444b43d318c92f360a13a25561a6a65b10c0631b8ed305a426dbaa6c380e"
+  "model.safetensors-00002-of-00004.safetensors 5335161512 31c7d7e2dd5d207840b31cc59083c8f4c4718959149e0358c0364052bb9a0330"
+  "model.safetensors-00003-of-00004.safetensors 5368717440 7ec36ba3a4176a44c3c0876ad80c56a2f70c84bf008d82e9501df642f17dadec"
+  "model.safetensors-00004-of-00004.safetensors 3325995712 b62b0c4cd7e44edee103ee8f4fe225f246d5e768e07bfd5f25b63a8aa1fdd0c6"
+  "model.safetensors.index.json 79657 -"
+  "config.json 3126 -"
+  "chat_template.jinja 7756 -"
+  "tokenizer.json 12807982 5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42"
+  "tokenizer_config.json 16710 -"
+  "vocab.json 6722759 -"
+  "merges.txt 3353259 -"
+  "preprocessor_config.json 390 -"
+  "video_preprocessor_config.json 385 -"
+)
+
+mkdir -p "$DEST"
+status=0
+for entry in "${FILES[@]}"; do
+  read -r name size sha <<<"$entry"
+  f="$DEST/$name"
+  have=0
+  if [[ -f "$f" ]] && [[ "$(stat -c%s "$f")" == "$size" ]]; then
+    if [[ "$sha" == "-" ]] || sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+      have=1
+    fi
+  fi
+  if (( have )); then
+    echo "OK        $name"
+    continue
+  fi
+  if (( VERIFY_ONLY )); then
+    echo "MISSING   $name"
+    status=1
+    continue
+  fi
+  echo "FETCHING  $name"
+  tmp="$f.part"
+  curl -fsSL --fail-with-body -o "$tmp" "$BASE/$name"
+  mv "$tmp" "$f"
+  if [[ "$(stat -c%s "$f")" != "$size" ]]; then
+    echo "SIZE FAIL $name ($(stat -c%s "$f") != $size)"
+    status=1
+    continue
+  fi
+  if [[ "$sha" != "-" ]] && ! sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+    echo "SHA FAIL  $name"
+    status=1
+    continue
+  fi
+  echo "VERIFIED  $name"
+done
+if (( status == 0 )); then
+  echo "DOWNLOAD_DONE $DEST"
+else
+  echo "DOWNLOAD_FAILED $DEST" >&2
+fi
+exit "$status"

old/download-frank-small-sglang-int8.sh

@@ -0,0 +1,93 @@
+#!/usr/bin/env bash
+# Download MiMo-V2.6-Distill-Qwen-9B-INT8-W8A16-AutoRound for frank-small
+# (SGLang, RTX 3090). Public repo — no auth header on the command line.
+#
+# Per-file FRESH downloads, no resume: `curl -C -` once produced a 213 MB
+# oversized file on a 31 GiB payload; a fresh -f fetch plus size+sha256 is
+# the house rule.
+#
+# Usage:
+#   scripts/download-frank-small.sh [DEST_DIR] [--verify-only]
+#
+# LFS shards are verified against the Hub LFS sha256; non-LFS text files
+# against their declared byte size. Idempotent: complete files are skipped.
+set -euo pipefail
+
+REPO="letechlead/MiMo-V2.6-Distill-Qwen-9B-INT8-W8A16-AutoRound"
+BASE="https://huggingface.co/${REPO}/resolve/main"
+
+DEST=""
+VERIFY_ONLY=0
+for arg in "$@"; do
+    case "$arg" in
+        --verify-only) VERIFY_ONLY=1 ;;
+        *) DEST="$arg" ;;
+    esac
+done
+DEST="${DEST:-/home/dank/models/MiMo-V2.6-Distill-Qwen-9B-INT8-W8A16-AutoRound}"
+
+# name bytes sha256(- = non-LFS, size-only check)
+FILES=(
+  "model-00001-of-00008.safetensors 2946296312 ddb6117c04cb6feaf7c265e5d75efadf21949037d73253a8e4521d40e164366d"
+  "model-00002-of-00008.safetensors 1402134792 2dc78d55d1d33d7b3b21dd6f8b4b817b62c2fb1376e6448b322428798c10e257"
+  "model-00003-of-00008.safetensors 1427365904 5c797d14e345309d67c78306b9de4aa01a21dbd92dd98e40b0ad69b2900e9203"
+  "model-00004-of-00008.safetensors 1420041856 7402f852e2c7a24ebf4694b98a220e71dfb2c93ba3776272cf2acfc65acd8126"
+  "model-00005-of-00008.safetensors 1410655648 991cbf2eda2ff91fc3ffbb15591d7fcac340fb10d9c33bd3da20010e2f3f4090"
+  "model-00006-of-00008.safetensors 1427627856 efc42448054cfb5e3061e154ba4539204088ce95176d689a95b58de885ef7eeb"
+  "model-00007-of-00008.safetensors 2035442408 75a101fc71ae6f2e1f5c4816f711d62cf1d7a0e02ef6e5182361b8c11e72b626"
+  "model-00008-of-00008.safetensors 822153664 0fb90cb50e2d336fbad58bd4c01880eeddbf537c1dca1cf0ae291195d5efbb4b"
+  "tokenizer.json 19989325 06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523"
+  "model.safetensors.index.json 106278 -"
+  "config.json 4608 -"
+  "generation_config.json 161 -"
+  "quantization_config.json 1654 -"
+  "chat_template.jinja 3916 -"
+  "tokenizer_config.json 1168 -"
+  "preprocessor_config.json 443 -"
+  "processor_config.json 1191 -"
+  "README.md 8448 -"
+)
+
+mkdir -p "$DEST"
+status=0
+for entry in "${FILES[@]}"; do
+  read -r name size sha <<<"$entry"
+  f="$DEST/$name"
+  have=0
+  if [[ -f "$f" ]] && [[ "$(stat -c%s "$f")" == "$size" ]]; then
+    if [[ "$sha" == "-" ]] || sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+      have=1
+    fi
+  fi
+  if (( have )); then
+    echo "OK        $name"
+    continue
+  fi
+  if (( VERIFY_ONLY )); then
+    echo "MISSING   $name"
+    status=1
+    continue
+  fi
+  echo "FETCHING  $name"
+  tmp="$f.part"
+  # no -C resume; -f + --fail-with-body so an HTTP error is not a success
+  curl -fsSL --fail-with-body -o "$tmp" "$BASE/$name"
+  mv "$tmp" "$f"
+  if [[ "$(stat -c%s "$f")" != "$size" ]]; then
+    echo "SIZE FAIL $name ($(stat -c%s "$f") != $size)"
+    status=1
+    continue
+  fi
+  if [[ "$sha" != "-" ]] && ! sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+    echo "SHA FAIL  $name"
+    status=1
+    continue
+  fi
+  echo "VERIFIED  $name"
+done
+if (( status == 0 )); then
+  echo "DOWNLOAD_DONE $DEST"
+else
+  echo "DOWNLOAD_FAILED $DEST" >&2
+fi
+exit "$status"

prose_prompt.txt

@@ -0,0 +1,86 @@
+# Subsurface Aquifer Monitoring Network - System Design Document (Revision 7)
+
+This document describes the deployment of a subsurface aquifer monitoring network for the Cascadia sedimentary basin. It is written for the operations crew and the regional water authority. Every sensor node, relay, aggregation service, calibration procedure, and failure mode below carries a stable alphanumeric identifier that appears in every table and in the field notes.
+
+## Sensor nodes
+Node SN-000-08B9 sits in a discharge lens at 359.0 m below datum and reports a salinity of 2.38 ppt with the fractured interval reading 5.40 degC at dawn and the wellhead porous 1.72 m3/d while the electrode pair returns 3157 uS/cm.
+Node SN-001-0A65 sits in a percolate lens at 25.9 m below datum and steps down to a salinity of 2.13 ppt with the diapir interval reading 8.55 degC at dawn and the wellhead silty 0.67 m3/d while the electrode pair returns 979 uS/cm.
+Node SN-002-0571 sits in a diapir lens at 287.8 m below datum and peaks at a salinity of 7.63 ppt with the karst interval reading 4.26 degC at dawn and the wellhead spooned 11.50 m3/d while the electrode pair returns 2987 uS/cm.
+Node SN-003-0CE9 sits in a confining lens at 169.5 m below datum and logs a salinity of 0.96 ppt with the alluvial interval reading 5.03 degC at dawn and the wellhead salinized 12.08 m3/d while the electrode pair returns 2607 uS/cm.
+Node SN-004-0E2A sits in a freshened lens at 70.4 m below datum and settles near a salinity of 8.68 ppt with the spooned interval reading 4.88 degC at dawn and the wellhead siphon 4.38 m3/d while the electrode pair returns 2707 uS/cm.
+Node SN-005-0DDD sits in a percolate lens at 44.6 m below datum and reports a salinity of 6.25 ppt with the alluvial interval reading 10.85 degC at dawn and the wellhead brackish 13.99 m3/d while the electrode pair returns 3618 uS/cm.
+Node SN-006-0EFC sits in a till lens at 309.5 m below datum and reports a salinity of 3.49 ppt with the lacustrine interval reading 7.38 degC at dawn and the wellhead spooned 3.21 m3/d while the electrode pair returns 1253 uS/cm.
+Node SN-007-0B02 sits in a recharge lens at 262.0 m below datum and steps down to a salinity of 2.38 ppt with the discharge interval reading 8.18 degC at dawn and the wellhead perched 4.05 m3/d while the electrode pair returns 3900 uS/cm.
+Node SN-008-0448 sits in a karst lens at 406.5 m below datum and reports a salinity of 7.32 ppt with the upconing interval reading 6.17 degC at dawn and the wellhead fluvial 0.75 m3/d while the electrode pair returns 1448 uS/cm.
+Node SN-009-043C sits in a alluvial lens at 159.3 m below datum and flags a salinity of 6.20 ppt with the percolate interval reading 7.60 degC at dawn and the wellhead brackish 13.01 m3/d while the electrode pair returns 2025 uS/cm.
+Node SN-010-0FF5 sits in a karst lens at 264.2 m below datum and flags a salinity of 7.05 ppt with the freshened interval reading 7.88 degC at dawn and the wellhead siphon 8.41 m3/d while the electrode pair returns 1635 uS/cm.
+Node SN-011-098A sits in a drain lens at 365.7 m below datum and reports a salinity of 8.11 ppt with the wedge interval reading 5.51 degC at dawn and the wellhead glacial 2.52 m3/d while the electrode pair returns 2916 uS/cm.
+Node SN-012-0AE8 sits in a till lens at 259.6 m below datum and steps down to a salinity of 9.13 ppt with the recharge interval reading 11.60 degC at dawn and the wellhead spooned 0.46 m3/d while the electrode pair returns 3077 uS/cm.
+Node SN-013-0B83 sits in a salinized lens at 171.2 m below datum and peaks at a salinity of 2.83 ppt with the lacustrine interval reading 5.56 degC at dawn and the wellhead discharge 0.35 m3/d while the electrode pair returns 3083 uS/cm.
+Node SN-014-032E sits in a drain lens at 249.6 m below datum and flags a salinity of 1.09 ppt with the salinized interval reading 9.58 degC at dawn and the wellhead perched 12.00 m3/d while the electrode pair returns 2221 uS/cm.
+Node SN-015-004F sits in a aeolian lens at 264.4 m below datum and reports a salinity of 7.34 ppt with the salinized interval reading 8.76 degC at dawn and the wellhead perched 0.31 m3/d while the electrode pair returns 1483 uS/cm.
+Node SN-016-07C6 sits in a aeolian lens at 155.9 m below datum and reports a salinity of 0.64 ppt with the mineralized interval reading 11.75 degC at dawn and the wellhead diluted 13.46 m3/d while the electrode pair returns 524 uS/cm.
+Node SN-017-06C5 sits in a salinized lens at 72.1 m below datum and averages a salinity of 4.52 ppt with the siphon interval reading 8.93 degC at dawn and the wellhead confining 3.98 m3/d while the electrode pair returns 3687 uS/cm.
+Node SN-018-0E64 sits in a siphon lens at 334.9 m below datum and steps down to a salinity of 6.73 ppt with the salinized interval reading 7.63 degC at dawn and the wellhead freshened 9.64 m3/d while the electrode pair returns 1681 uS/cm.
+Node SN-019-0122 sits in a spooned lens at 117.2 m below datum and logs a salinity of 3.24 ppt with the fractured interval reading 9.26 degC at dawn and the wellhead karst 3.50 m3/d while the electrode pair returns 1065 uS/cm.
+Node SN-020-021D sits in a karst lens at 414.3 m below datum and recovers to a salinity of 0.76 ppt with the glacial interval reading 6.25 degC at dawn and the wellhead silty 9.60 m3/d while the electrode pair returns 1041 uS/cm.
+Node SN-021-06C6 sits in a percolate lens at 379.2 m below datum and drifts to a salinity of 7.61 ppt with the wedge interval reading 5.84 degC at dawn and the wellhead karst 1.65 m3/d while the electrode pair returns 1913 uS/cm.
+Node SN-022-0DC9 sits in a spooned lens at 327.1 m below datum and reports a salinity of 9.25 ppt with the freshened interval reading 5.05 degC at dawn and the wellhead silty 5.90 m3/d while the electrode pair returns 1544 uS/cm.
+Node SN-023-0717 sits in a karst lens at 222.0 m below datum and steps down to a salinity of 4.02 ppt with the alluvial interval reading 6.60 degC at dawn and the wellhead siphon 3.77 m3/d while the electrode pair returns 3892 uS/cm.
+Node SN-024-02A9 sits in a fractured lens at 479.7 m below datum and drifts to a salinity of 7.88 ppt with the silty interval reading 12.53 degC at dawn and the wellhead discharge 13.18 m3/d while the electrode pair returns 3592 uS/cm.
+Node SN-025-0C8F sits in a wedge lens at 434.3 m below datum and holds a salinity of 1.63 ppt with the alluvial interval reading 4.22 degC at dawn and the wellhead till 5.73 m3/d while the electrode pair returns 3905 uS/cm.
+Node SN-026-04BF sits in a colluvial lens at 460.0 m below datum and averages a salinity of 9.36 ppt with the diapir interval reading 8.98 degC at dawn and the wellhead brackish 10.29 m3/d while the electrode pair returns 802 uS/cm.
+Node SN-027-0DA4 sits in a silty lens at 265.7 m below datum and steps down to a salinity of 7.06 ppt with the percolate interval reading 4.69 degC at dawn and the wellhead freshened 8.42 m3/d while the electrode pair returns 2201 uS/cm.
+Node SN-028-01EB sits in a perched lens at 49.5 m below datum and recovers to a salinity of 1.83 ppt with the silty interval reading 9.32 degC at dawn and the wellhead drain 9.69 m3/d while the electrode pair returns 1126 uS/cm.
+Node SN-029-042C sits in a karst lens at 30.6 m below datum and averages a salinity of 0.86 ppt with the percolate interval reading 9.85 degC at dawn and the wellhead recharge 8.16 m3/d while the electrode pair returns 1452 uS/cm.
+
+## Relays
+Relay RL-00-02181 reburses the upconing trunk from the brackish margin and peaks at a round-trip time of 131.8 ms with a frame loss of 1.04 percent across the night window.
+Relay RL-01-00262 reburses the spooned trunk from the fluvial margin and holds a round-trip time of 372.2 ms with a frame loss of 3.65 percent across the night window.
+Relay RL-02-03691 reburses the recharge trunk from the percolate margin and steps down to a round-trip time of 398.5 ms with a frame loss of 0.29 percent across the night window.
+Relay RL-03-0E117 reburses the lacustrine trunk from the confining margin and averages a round-trip time of 373.9 ms with a frame loss of 3.44 percent across the night window.
+Relay RL-04-0B41B reburses the aeolian trunk from the colluvial margin and peaks at a round-trip time of 70.7 ms with a frame loss of 3.25 percent across the night window.
+Relay RL-05-0D12D reburses the recharge trunk from the diluted margin and steps down to a round-trip time of 264.7 ms with a frame loss of 0.03 percent across the night window.
+Relay RL-06-043B4 reburses the colluvial trunk from the upconing margin and reports a round-trip time of 49.5 ms with a frame loss of 3.42 percent across the night window.
+Relay RL-07-09AD6 reburses the fractured trunk from the freshened margin and averages a round-trip time of 225.3 ms with a frame loss of 1.06 percent across the night window.
+Relay RL-08-04394 reburses the brackish trunk from the fluvial margin and steps down to a round-trip time of 88.6 ms with a frame loss of 2.47 percent across the night window.
+Relay RL-09-017A0 reburses the wedge trunk from the lacustrine margin and drifts to a round-trip time of 363.0 ms with a frame loss of 3.30 percent across the night window.
+Relay RL-10-0217D reburses the mineralized trunk from the lacustrine margin and peaks at a round-trip time of 26.2 ms with a frame loss of 1.30 percent across the night window.
+Relay RL-11-08F97 reburses the perched trunk from the freshened margin and logs a round-trip time of 181.8 ms with a frame loss of 2.75 percent across the night window.
+Relay RL-12-02628 reburses the fractured trunk from the glacial margin and steps down to a round-trip time of 378.5 ms with a frame loss of 2.69 percent across the night window.
+Relay RL-13-06E06 reburses the silty trunk from the mineralized margin and flags a round-trip time of 153.4 ms with a frame loss of 2.15 percent across the night window.
+
+## Aggregation services
+Aggregator AG-00-3BB logs a queue depth of 5.2 messages and a histogram bucket of 6.82 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-01-0FF reports a queue depth of 86.4 messages and a histogram bucket of 4.10 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-02-37F drifts to a queue depth of 34.7 messages and a histogram bucket of 10.25 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-03-3B3 averages a queue depth of 93.5 messages and a histogram bucket of 13.59 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-04-33E flags a queue depth of 83.2 messages and a histogram bucket of 17.58 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-05-2F2 settles near a queue depth of 84.8 messages and a histogram bucket of 0.84 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-06-2AD averages a queue depth of 75.4 messages and a histogram bucket of 7.65 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-07-06E drifts to a queue depth of 26.5 messages and a histogram bucket of 14.26 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-08-0E3 averages a queue depth of 83.9 messages and a histogram bucket of 15.51 seconds before the dedupe pass settles the stream for the next rotation.
+Aggregator AG-09-138 averages a queue depth of 78.6 messages and a histogram bucket of 8.50 seconds before the dedupe pass settles the stream for the next rotation.
+
+## Calibration procedures
+Procedure CP-00-047 steps the probe through a two-point rinse, then applies a trim of 0.2084 at the 66 milli-bar mark and records a residual of 0.02052 before the next field seal is crimped.
+Procedure CP-01-0A4 steps the probe through a two-point rinse, then applies a trim of 0.7793 at the 96 milli-bar mark and records a residual of 0.01468 before the next field seal is crimped.
+Procedure CP-02-054 steps the probe through a two-point rinse, then applies a trim of 0.4634 at the 68 milli-bar mark and records a residual of 0.04234 before the next field seal is crimped.
+Procedure CP-03-02D steps the probe through a two-point rinse, then applies a trim of 0.8459 at the 12 milli-bar mark and records a residual of 0.04855 before the next field seal is crimped.
+Procedure CP-04-098 steps the probe through a two-point rinse, then applies a trim of 0.5267 at the 97 milli-bar mark and records a residual of 0.00287 before the next field seal is crimped.
+Procedure CP-05-09B steps the probe through a two-point rinse, then applies a trim of 0.3968 at the 72 milli-bar mark and records a residual of 0.01637 before the next field seal is crimped.
+Procedure CP-06-030 steps the probe through a two-point rinse, then applies a trim of 0.8860 at the 12 milli-bar mark and records a residual of 0.04508 before the next field seal is crimped.
+Procedure CP-07-0EC steps the probe through a two-point rinse, then applies a trim of 0.2367 at the 70 milli-bar mark and records a residual of 0.00108 before the next field seal is crimped.
+Procedure CP-08-0BC steps the probe through a two-point rinse, then applies a trim of 0.7278 at the 68 milli-bar mark and records a residual of 0.04707 before the next field seal is crimped.
+Procedure CP-09-0AA steps the probe through a two-point rinse, then applies a trim of 0.6069 at the 37 milli-bar mark and records a residual of 0.00443 before the next field seal is crimped.
+
+## Failure modes
+Mode FM-00-27 declares a fluvial excursion when two independent probes disagree beyond the band for three rotations; the fleet reports a mean time between failures of 36902 hours for this class of excursion.
+Mode FM-01-13 declares a drain excursion when two independent probes disagree beyond the band for three rotations; the fleet peaks at a mean time between failures of 29042 hours for this class of excursion.
+Mode FM-02-14 declares a till excursion when two independent probes disagree beyond the band for three rotations; the fleet peaks at a mean time between failures of 26976 hours for this class of excursion.
+Mode FM-03-0C declares a diapir excursion when two independent probes disagree beyond the band for three rotations; the fleet drifts to a mean time between failures of 22578 hours for this class of excursion.
+Mode FM-04-39 declares a perched excursion when two independent probes disagree beyond the band for three rotations; the fleet recovers to a mean time between failures of 27383 hours for this class of excursion.
+Mode FM-05-19 declares a siphon excursion when two independent probes disagree beyond the band for three rotations; the fleet logs a mean time between failures of 23152 hours for this class of excursion.
+
+## End of document
+Summarize the network's dominant risk in one paragraph, then list the three node identifiers whose readings most limit the model this month.

prose_request.json

@@ -0,0 +1 @@
+{"model": "frank-large", "prompt": "# Subsurface Aquifer Monitoring Network - System Design Document (Revision 7)\n\nThis document describes the deployment of a subsurface aquifer monitoring network for the Cascadia sedimentary basin. It is written for the operations crew and the regional water authority. Every sensor node, relay, aggregation service, calibration procedure, and failure mode below carries a stable alphanumeric identifier that appears in every table and in the field notes.\n\n## Sensor nodes\nNode SN-000-08B9 sits in a discharge lens at 359.0 m below datum and reports a salinity of 2.38 ppt with the fractured interval reading 5.40 degC at dawn and the wellhead porous 1.72 m3/d while the electrode pair returns 3157 uS/cm.\nNode SN-001-0A65 sits in a percolate lens at 25.9 m below datum and steps down to a salinity of 2.13 ppt with the diapir interval reading 8.55 degC at dawn and the wellhead silty 0.67 m3/d while the electrode pair returns 979 uS/cm.\nNode SN-002-0571 sits in a diapir lens at 287.8 m below datum and peaks at a salinity of 7.63 ppt with the karst interval reading 4.26 degC at dawn and the wellhead spooned 11.50 m3/d while the electrode pair returns 2987 uS/cm.\nNode SN-003-0CE9 sits in a confining lens at 169.5 m below datum and logs a salinity of 0.96 ppt with the alluvial interval reading 5.03 degC at dawn and the wellhead salinized 12.08 m3/d while the electrode pair returns 2607 uS/cm.\nNode SN-004-0E2A sits in a freshened lens at 70.4 m below datum and settles near a salinity of 8.68 ppt with the spooned interval reading 4.88 degC at dawn and the wellhead siphon 4.38 m3/d while the electrode pair returns 2707 uS/cm.\nNode SN-005-0DDD sits in a percolate lens at 44.6 m below datum and reports a salinity of 6.25 ppt with the alluvial interval reading 10.85 degC at dawn and the wellhead brackish 13.99 m3/d while the electrode pair returns 3618 uS/cm.\nNode SN-006-0EFC sits in a till lens at 309.5 m below datum and reports a salinity of 3.49 ppt with the lacustrine interval reading 7.38 degC at dawn and the wellhead spooned 3.21 m3/d while the electrode pair returns 1253 uS/cm.\nNode SN-007-0B02 sits in a recharge lens at 262.0 m below datum and steps down to a salinity of 2.38 ppt with the discharge interval reading 8.18 degC at dawn and the wellhead perched 4.05 m3/d while the electrode pair returns 3900 uS/cm.\nNode SN-008-0448 sits in a karst lens at 406.5 m below datum and reports a salinity of 7.32 ppt with the upconing interval reading 6.17 degC at dawn and the wellhead fluvial 0.75 m3/d while the electrode pair returns 1448 uS/cm.\nNode SN-009-043C sits in a alluvial lens at 159.3 m below datum and flags a salinity of 6.20 ppt with the percolate interval reading 7.60 degC at dawn and the wellhead brackish 13.01 m3/d while the electrode pair returns 2025 uS/cm.\nNode SN-010-0FF5 sits in a karst lens at 264.2 m below datum and flags a salinity of 7.05 ppt with the freshened interval reading 7.88 degC at dawn and the wellhead siphon 8.41 m3/d while the electrode pair returns 1635 uS/cm.\nNode SN-011-098A sits in a drain lens at 365.7 m below datum and reports a salinity of 8.11 ppt with the wedge interval reading 5.51 degC at dawn and the wellhead glacial 2.52 m3/d while the electrode pair returns 2916 uS/cm.\nNode SN-012-0AE8 sits in a till lens at 259.6 m below datum and steps down to a salinity of 9.13 ppt with the recharge interval reading 11.60 degC at dawn and the wellhead spooned 0.46 m3/d while the electrode pair returns 3077 uS/cm.\nNode SN-013-0B83 sits in a salinized lens at 171.2 m below datum and peaks at a salinity of 2.83 ppt with the lacustrine interval reading 5.56 degC at dawn and the wellhead discharge 0.35 m3/d while the electrode pair returns 3083 uS/cm.\nNode SN-014-032E sits in a drain lens at 249.6 m below datum and flags a salinity of 1.09 ppt with the salinized interval reading 9.58 degC at dawn and the wellhead perched 12.00 m3/d while the electrode pair returns 2221 uS/cm.\nNode SN-015-004F sits in a aeolian lens at 264.4 m below datum and reports a salinity of 7.34 ppt with the salinized interval reading 8.76 degC at dawn and the wellhead perched 0.31 m3/d while the electrode pair returns 1483 uS/cm.\nNode SN-016-07C6 sits in a aeolian lens at 155.9 m below datum and reports a salinity of 0.64 ppt with the mineralized interval reading 11.75 degC at dawn and the wellhead diluted 13.46 m3/d while the electrode pair returns 524 uS/cm.\nNode SN-017-06C5 sits in a salinized lens at 72.1 m below datum and averages a salinity of 4.52 ppt with the siphon interval reading 8.93 degC at dawn and the wellhead confining 3.98 m3/d while the electrode pair returns 3687 uS/cm.\nNode SN-018-0E64 sits in a siphon lens at 334.9 m below datum and steps down to a salinity of 6.73 ppt with the salinized interval reading 7.63 degC at dawn and the wellhead freshened 9.64 m3/d while the electrode pair returns 1681 uS/cm.\nNode SN-019-0122 sits in a spooned lens at 1Line truncated

scripts/autoresearch-bench.py

@@ -0,0 +1,141 @@
+#!/usr/bin/env python3
+"""autoresearch-bench.py — deterministic greedy battery for frank-large.
+
+Runs ON Dankclaw (no LAN jitter) against the candidate container on :7770.
+Fixed embedded corpus, temperature 0, streaming so decode speed is measured
+between real token arrivals (not request totals). Median of REPS samples per
+metric is the reported value; medians are far more stable than means under
+GPU thermal drift.
+
+Metrics emitted (stdout, one per line):
+  METRIC decode_tokens_per_sec=<float>   primary, higher better
+  METRIC prefill_tokens_per_sec=<float>  secondary
+  METRIC ttft_ms=<float>                 secondary, lower better
+Exit non-zero if the endpoint is unhealthy or the served alias is wrong.
+"""
+import argparse
+import json
+import statistics
+import sys
+import time
+import urllib.request
+
+# Fixed corpus: deterministic, ~28k tokens after tokenization (same scale as
+# the 2026-10-01 bench window). Built by repeating a fixed paragraph block a
+# fixed number of times - NO random, NO clock, NO network.
+_BLOCK = (
+    "The rendering pipeline walks the scene graph once per frame, resolving "
+    "material bindings into descriptor sets before any draw call is recorded. "
+    "Because the allocator recycles command buffers only after their fence "
+    "signals, a submission that outlives its frame must pin the pool entry; "
+    "otherwise the next frame overwrites recorded commands in flight. The "
+    "validation layer flags exactly this class of lifetime bug, but only for "
+    "buffers it can see - GPU-side timestamps and queue-family ownership "
+    "transfers hide in the driver's own bookkeeping, which is why release "
+    "builds must lean on explicit barriers and why a frame-time regression "
+    "often shows up as corruption one frame later than its cause.\n"
+)
+LONG_CTX = _BLOCK * 220  # ~28k-token prefill, matching the deployed workload
+_LONG = LONG_CTX + "\nNow, in one focused paragraph (80-150 words), explain " \
+    "which lifetime bugs this pipeline can still suffer and what explicit " \
+    "mitigations the text implies."
+_SHORT = "Design a rate limiter for a homelab reverse proxy that must not " \
+    "lock out the LAN. Think it through, then give the algorithm."
+
+REPS = 3
+MAX_TOKENS = 256
+REQ_TIMEOUT = 900
+
+
+def stream_completion(url: str, model: str, prompt: str):
+    """Greedy streaming completion.
+
+    Returns (ttft_s, decode_tps, prefill_tps, prompt_tokens, n_tokens).
+    ttft is RELATIVE to request send: time.perf_counter()'s epoch is host
+    boot, and returning the raw stamp read as an 11-day first token.
+    """
+    body = json.dumps({
+        "model": model,
+        "messages": [{"role": "user", "content": prompt}],
+        "max_tokens": MAX_TOKENS,
+        "temperature": 0,
+        "stream": True,
+        "stream_options": {"include_usage": True},
+    }).encode()
+    req = urllib.request.Request(
+        url.rstrip("/") + "/v1/chat/completions", data=body,
+        headers={"Content-Type": "application/json"})
+    t0 = time.perf_counter()
+    first = None
+    stamps = []
+    usage = {}
+    with urllib.request.urlopen(req, timeout=REQ_TIMEOUT) as resp:
+        for raw in resp:
+            line = raw.decode(errors="replace").strip()
+            if not line.startswith("data:"):
+                continue
+            payload = line[5:].strip()
+            if payload == "[DONE]":
+                break
+            chunk = json.loads(payload)
+            if chunk.get("usage"):
+                usage = chunk["usage"]
+            for ch in chunk.get("choices", []):
+                d = ch.get("delta") or {}
+                # reasoning AND content deltas both arrive token by token;
+                # count either so greedy text length drives the rate.
+                # (llama.cpp names the thinking field reasoning_content.)
+                if d.get("content") or d.get("reasoning") or d.get("reasoning_content"):
+                    now = time.perf_counter()
+                    if first is None:
+                        first = now
+                    stamps.append(now)
+    if first is None or len(stamps) < 2:
+        # A zero-argmax degenerate answer would poison the median; raise.
+        raise RuntimeError(f"stream produced {len(stamps)} tokens")
+    ttft_s = first - t0
+    decode_s = stamps[-1] - stamps[0]
+    decode_tps = (len(stamps) - 1) / decode_s if decode_s > 0 else 0.0
+    prompt_tokens = usage.get("prompt_tokens") or 0
+    prefill_tps = prompt_tokens / ttft_s if prompt_tokens else 0.0
+    return ttft_s, decode_tps, prefill_tps, prompt_tokens, len(stamps)
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--url", default="http://127.0.0.1:7770")
+    ap.add_argument("--model", default="frank-large")
+    args = ap.parse_args()
+
+    with urllib.request.urlopen(args.url.rstrip("/") + "/v1/models",
+                                timeout=10) as r:
+        ids = [m["id"] for m in json.load(r)["data"]]
+    if args.model not in ids:
+        print(f"FATAL: served ids {ids}, expected {args.model}",
+              file=sys.stderr)
+        sys.exit(2)
+
+    # Warmup (compile/graph warm + allocator settle), NOT measured.
+    stream_completion(args.url, args.model, _SHORT)
+
+    decodes, prefills, ttfts = [], [], []
+    for rep in range(REPS):
+        for name, prompt in (("long", _LONG), ("short", _SHORT)):
+            ttft, dt, pt, ptk, ntok = stream_completion(
+                args.url, args.model, prompt)
+            decodes.append(dt)
+            ttfts.append(ttft * 1000.0)
+            if name == "long":
+                prefills.append(pt)
+            print(f"  rep{rep} {name:5s} ttft={ttft*1000:8.1f}ms "
+                  f"prefill={pt:8.1f} t/s decode={dt:6.2f} t/s "
+                  f"({ntok} tok, prompt {ptk})", file=sys.stderr)
+
+    med = lambda xs: statistics.median(xs)
+    print(f"METRIC decode_tokens_per_sec={med(decodes):.2f}")
+    print(f"METRIC prefill_tokens_per_sec={med(prefills):.1f}")
+    print(f"METRIC ttft_ms={med(ttfts):.1f}")
+
+
+if __name__ == "__main__":
+    main()

scripts/autoresearch/frank-primary-arms.sh

@@ -0,0 +1,176 @@
+#!/usr/bin/env bash
+# frank-primary-arms.sh — run a SEQUENCE of frank-primary bench arms and keep
+# every arm's evidence. Runs on the operator box; each arm is one full window
+# of ./autoresearch.sh (stop+mask -> boot the arm as the service -> measure ->
+# restore production and verify), so the fleet is only ever down for one arm at
+# a time and the production unit is back up between arms.
+#
+# An arm is the repo's own frank-primary.env plus an OVERLAY of engine knobs
+# appended after it (later assignment wins when the serve script sources it),
+# staged through FRANK_PRIMARY_AR_ENV_FILE so the checkout is never mutated and
+# each arm gets its own fingerprint (the memo can never answer one arm with
+# another arm's numbers).
+#
+# Usage:  scripts/autoresearch/frank-primary-arms.sh [arm ...]
+#         scripts/autoresearch/frank-primary-arms.sh --list
+# Output, under /tmp/frank-primary-arms-<UTC stamp>/ (override with ARMS_DIR):
+#
+#   <arm>.out        full autoresearch output (metrics + engine evidence)
+#   <arm>.env        the exact staged env
+#   <arm>.rc         window exit code
+#   summary.tsv      arm, rc, decode_tps, decode_tps_t0.7, kv_tokens, boot_s
+#
+# Exit 0 when every arm produced metrics; 1 if any arm failed (a failed boot is
+# a RESULT — the window still restored production, and the rc says which arm).
+set -uo pipefail
+
+SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
+REPO="$(cd -- "$SCRIPT_DIR/../.." && pwd)"
+BASE_ENV="$REPO/scripts/frank-primary.env"
+ARMS_DIR="${ARMS_DIR:-/tmp/frank-primary-arms-$(date -u +%Y%m%dT%H%M%SZ)}"
+
+# --- arms ------------------------------------------------------------------
+# baseline: the shipped shape E, measured with THIS harness so the comparison is
+# matched (the ~29.6 t/s figure in the docs is an older shape, and no shape-E
+# decode number exists yet).
+OVER_baseline=()
+# B: MTP3 with the vendor's probabilistic draft sampling, a capture ladder whose
+# step is num_speculative_tokens + 1 = 4, and the sampling override. Precise
+# memory stays ON — this arm is simultaneously the MTP test and the test of
+# whether MTP's drafter workspace and the fitted 5.25 GiB overhead can coexist.
+SPEC3='{"method":"mtp","num_speculative_tokens":3,"draft_sample_method":"probabilistic"}'
+LADDER3='{"cudagraph_capture_sizes":[4,8,12],"max_cudagraph_capture_size":12,"inductor_compile_config":{"combo_kernels":false,"benchmark_combo_kernel":false}}'
+# The value must arrive in the staged env file SINGLE-QUOTED: sourcing an env
+# file performs quote removal, so an unquoted {"temperature":0.7} would reach
+# the engine as {temperature:0.7}.
+GEN07='{"temperature":0.7}'
+OVER_B=(
+    "FRANK_PRIMARY_SPEC='$SPEC3'"
+    "FRANK_PRIMARY_COMPCONFIG='$LADDER3'"
+    'FRANK_PRIMARY_GEN_CONFIG='"'$GEN07'"
+)
+# B2: identical to B with the default (7.60 GiB) overhead, i.e. precise memory
+# OFF. It is the fallback if B dies at graph capture, and it prices the arena
+# that precise memory buys (7,522 slots -> ~6,496).
+OVER_B2=(
+    "FRANK_PRIMARY_SPEC='$SPEC3'"
+    "FRANK_PRIMARY_COMPCONFIG='$LADDER3'"
+    'FRANK_PRIMARY_GEN_CONFIG='"'$GEN07'"
+    'FRANK_PRIMARY_PRECISE_MEMORY='
+)
+# C: the vendor's concurrency shape — SEQS 16 with the [4..32] ladder and util
+# 0.95. This is also the direct test of the theory that pinning the capture
+# ladder unlocks SEQS > 3 (the SEQS=2 failure was query_start_loc specialised to
+# the constant 3, which is NOT in the default ladder; 4 is).
+LADDER16='{"cudagraph_capture_sizes":[4,8,12,16,20,24,28,32],"max_cudagraph_capture_size":32,"inductor_compile_config":{"combo_kernels":false,"benchmark_combo_kernel":false}}'
+OVER_C=(
+    "FRANK_PRIMARY_SPEC='$SPEC3'"
+    "FRANK_PRIMARY_COMPCONFIG='$LADDER16'"
+    'FRANK_PRIMARY_GEN_CONFIG='"'$GEN07'"
+    'FRANK_PRIMARY_SEQS=16'
+    'FRANK_PRIMARY_MNBT=4096'
+    'FRANK_PRIMARY_UTIL=0.95'
+)
+
+# D/E: the CONTEXT arms. Measured 2026-10-02 (arms run 20261002T233855Z): the
+# KV pool is a hard byte request divided by a PER-TOKEN PRICE, and MTP raises
+# that price because the drafter keeps its own KV per position:
+#   no spec: 3.15e9 / 398,658 tokens =  7,901 B/token  (1.99x of 200,704)
+#   MTP3:    3.15e9 / 313,002 tokens = 10,064 B/token  (1.56x)
+# So arm C lost a fifth of the context to the drafter, not to util. Frank's
+# contract is >=401,400 tokens over >=2 seqs = 2.00x of 200,704, which at
+# 10,064 B/token costs 4.04e9 -> ask 4.05e9 (+0.83 GiB/rank).
+# D is arm B with EXACTLY ONE variable changed (the KV bytes), so its result
+# says whether the 4k-depth collapse (B 23.09 vs baseline 33.99) was KV or the
+# batch shape.
+KV401K='FRANK_PRIMARY_KV_BYTES=4050000000'
+OVER_D=(
+    "FRANK_PRIMARY_SPEC='$SPEC3'"
+    "FRANK_PRIMARY_COMPCONFIG='$LADDER3'"
+    'FRANK_PRIMARY_GEN_CONFIG='"'$GEN07'"
+    "$KV401K"
+)
+# E: D with C's max_num_batched_tokens (4096 — C is the only arm that did NOT
+# collapse at 4k depth: 33.15 where B gave 23.09) and util 0.99, the value
+# Frank asked about. 0.99 is affordable on paper — production at util 0.98
+# measures 30.37/34.21 GiB per card, 3.58 GiB physically unused — but it leaves
+# only 0.32 GiB above the allowance while the CUDA-graph cost is an ANALYTIC
+# estimate (C's was 1.47 GiB over 24 graphs), so it is measured, not assumed.
+OVER_E=(
+    "FRANK_PRIMARY_SPEC='$SPEC3'"
+    "FRANK_PRIMARY_COMPCONFIG='$LADDER3'"
+    'FRANK_PRIMARY_GEN_CONFIG='"'$GEN07'"
+    "$KV401K"
+    'FRANK_PRIMARY_MNBT=4096'
+    'FRANK_PRIMARY_UTIL=0.99'
+)
+
+arm_overlay() {
+    case "$1" in
+        baseline) ;;                       # the shipped env, no overlay
+        B)  printf '%s\n' "${OVER_B[@]}" ;;
+        B2) printf '%s\n' "${OVER_B2[@]}" ;;
+        C)  printf '%s\n' "${OVER_C[@]}" ;;
+        D)  printf '%s\n' "${OVER_D[@]}" ;;
+        E)  printf '%s\n' "${OVER_E[@]}" ;;
+        *) return 1 ;;
+    esac
+}
+if [ "${1:-}" = "--list" ]; then
+    printf 'baseline B B2 C D E\n'
+    exit 0
+fi
+
+[ -f "$BASE_ENV" ] || { echo "[arms] missing $BASE_ENV" >&2; exit 1; }
+if [ "$#" -gt 0 ]; then ARMS=("$@"); else ARMS=(baseline B B2 C); fi
+mkdir -p "$ARMS_DIR"
+echo "[arms] dir $ARMS_DIR  arms: ${ARMS[*]}" >&2
+
+: >"$ARMS_DIR/summary.tsv"
+printf 'arm\trc\tdecode_tps\tdecode_tps_t0.7\tkv_tokens\tboot_s\n' >>"$ARMS_DIR/summary.tsv"
+overall=0
+
+for arm in "${ARMS[@]}"; do
+    if ! arm_overlay "$arm" >"$ARMS_DIR/$arm.overlay"; then
+        echo "[arms] unknown arm: $arm" >&2
+        overall=1
+        continue
+    fi
+    { cat "$BASE_ENV"; echo; echo "# --- arm $arm overlay ($(date -u +%FT%TZ)) ---"
+      arm_overlay "$arm"; } >"$ARMS_DIR/$arm.env"
+    # Gate the staged env BEFORE burning a ~20-minute window: every knob that
+    # holds JSON must still be one valid JSON object after the file is SOURCED
+    # (that is the step that silently eats unquoted braces).
+    if ! bash -c 'set -a; . "$1" >/dev/null; set +a
+        for k in FRANK_PRIMARY_SPEC FRANK_PRIMARY_COMPCONFIG FRANK_PRIMARY_GEN_CONFIG; do
+            v="${!k:-}"
+            [ -n "$v" ] || continue
+            case "$v" in
+                "{"*) python3 -c "import json,sys; json.loads(sys.argv[1])" "$v" \
+                        || { echo "[arms] $k is not valid JSON: $v" >&2; exit 1; } ;;
+                *) echo "[arms] $k is not a JSON object: $v" >&2; exit 1 ;;
+            esac
+        done' _ "$ARMS_DIR/$arm.env"; then
+        echo "[arms] $arm: staged env rejected, window not opened" >&2
+        echo 99 >"$ARMS_DIR/$arm.rc"
+        overall=1
+        continue
+    fi
+
+    echo "[arms] === $arm: opening window $(date -u +%H:%M:%S)Z ===" >&2
+    FRANK_PRIMARY_AR_ENV_FILE="$ARMS_DIR/$arm.env" \
+        "$REPO/autoresearch.sh" >"$ARMS_DIR/$arm.out" 2>&1
+    rc=$?
+    echo "$rc" >"$ARMS_DIR/$arm.rc"
+    echo "[arms] === $arm: rc=$rc $(date -u +%H:%M:%S)Z ===" >&2
+    grep -E '^METRIC ' "$ARMS_DIR/$arm.out" | sed 's/^METRIC //' >"$ARMS_DIR/$arm.metrics" || true
+    metric() { local v; v="$(sed -n "s/^$1=//p" "$ARMS_DIR/$arm.metrics" | tail -1)"; printf '%s' "${v:-$2}"; }
+    printf '%s\t%s\t%s\t%s\t%s\t%s\n' "$arm" "$rc" \
+        "$(metric decode_tps FAIL)" "$(metric decode_tps_t0.7 FAIL)" \
+        "$(metric kv_tokens ?)" "$(metric boot_seconds ?)" >>"$ARMS_DIR/summary.tsv"
+    [ "$rc" -ne 0 ] && overall=1
+done
+
+echo "[arms] summary:" >&2
+cat "$ARMS_DIR/summary.tsv" >&2
+exit "$overall"

scripts/autoresearch/frank-primary-measure.py

@@ -0,0 +1,216 @@
+#!/usr/bin/env python3
+"""frank-primary decode measurement — runs INSIDE the engine container.
+
+The method is the fleet's established one (the 2026-09-29 / 2026-10-01
+tuning campaigns, /home/dank/mtp_measure.py): SSE chunk counting is invalid
+under speculative decoding because rejected drafts change the accounting, so
+decode is measured from the server's own usage counters:
+
+    decode_tok_s = completion_tokens / (wall - ttft)
+
+with ttft = wall time to the FIRST content-bearing chunk (thinking arrives in
+`reasoning`/`delta.reasoning` on this stack and is excluded from the decode
+window — and is deterministic per prompt at temperature 0, so the exclusion
+cancels in any A/B on the same workload).
+
+Workload (fixed, deterministic, offline):
+  * prompts built from a large fixed-seed shuffled word list so tokens rarely
+    repeat (no n-gram/prefix predictability of the text itself), sized with
+    the server's own /v1/messages/count_tokens counter
+  * depths 115 / 1056 / 4096 target prompt tokens
+  * one 64-token warm pass per depth (warms the expert arena), then
+    max_tokens=256, 3 runs, median — repeated for every temperature in AR_TEMPS
+    (default 0 and 0.7). The prompt workload is deterministic (fixed seed, no
+    network, no time-of-day dependence); at temperature > 0 the generated text
+    is not, but the measured window is bounded by max_tokens, so the rate stays
+    comparable and the non-determinism is exactly the sampling that a
+    speculative-decode arm must survive.
+
+Env:
+  AR_BASE             endpoint (default http://127.0.0.1:7770)
+  AR_MODEL            served name (default frank-primary)
+  AR_PROMPT_CEILING   hard prompt ceiling in tokens (default 100000)
+  AR_TEMPS            comma-separated sampling temperatures (default "0,0.7")
+
+Output: progress on stderr, ONE json object on stdout (the last line).
+Exit codes: 0 ok, 2 endpoint unhealthy / no usable completion, 3 bad usage.
+"""
+import json
+import os
+import random
+import statistics
+import sys
+import time
+import urllib.error
+import urllib.request
+
+BASE = os.environ.get("AR_BASE", "http://127.0.0.1:7770")
+MODEL = os.environ.get("AR_MODEL", "frank-primary")
+PROMPT_CEILING = int(os.environ.get("AR_PROMPT_CEILING", "100000"))
+DEPTHS = (115, 1056, 4096)
+WARM_TOKENS = 64
+MAX_TOKENS = 256
+RUNS = 3
+# Temperature sweep. The unsuffixed keys are temperature 0 — byte-comparable to
+# every historical measurement — and AR_TEMPS adds more. The second one matters
+# for speculative decode: draft ACCEPTANCE is sampling-dependent (a greedy draft
+# matches a greedy target exactly and is rejected more under a non-greedy
+# target), so an MTP arm judged only at temp 0 would be judged on half the
+# contract. Production traffic is non-zero temperature (the payload's
+# generation_config says 1.0; an override to 0.7 is under test).
+TEMPS = tuple(float(x) for x in os.environ.get("AR_TEMPS", "0,0.7").split(","))
+
+# 43-word x 12 = 516-word vocabulary, seed 1234 — the campaign's exact list
+# and seed, so prompts are byte-identical to the historical measurements.
+_VOCAB = ("alpha bravo charlie delta echo foxtrot golf hotel india juliet kilo lima "
+          "mike november oscar papa quebec romeo sierra tango uniform victor whiskey "
+          "xray yankee zulu cardinal ordinal topology kernel scheduler resident "
+          "unified aperture expert arena slab offload prefetch gemm triton quant "
+          "tensor shard checkpoint manifest embed router cache page table window "
+          "gradient sparse dense moe attention head dim block stream decode prefill "
+          "bandwidth latency throughput residency eviction tier swap pin memlock").split()
+_R = random.Random(1234)
+_ORDER = list(range(len(_VOCAB)))
+_R.shuffle(_ORDER)
+
+
+def log(msg):
+    print("[measure] %s" % msg, file=sys.stderr, flush=True)
+
+
+def post_json(path, body, timeout):
+    req = urllib.request.Request(
+        BASE + path, data=json.dumps(body).encode(),
+        headers={"Content-Type": "application/json"})
+    with urllib.request.urlopen(req, timeout=timeout) as r:
+        return json.load(r)
+
+
+def count_tokens(text):
+    """Token count without a prefill, from the server's own counter."""
+    body = {"model": MODEL,
+            "messages": [{"role": "user", "content": text}]}
+    r = post_json("/v1/messages/count_tokens", body, timeout=300)
+    n = r.get("input_tokens") or r.get("count") or r.get("token_count")
+    if not n:
+        raise RuntimeError("count_tokens returned no count: %r" % (r,))
+    return int(n)
+
+
+def words(n):
+    return " ".join(_VOCAB[_ORDER[i % len(_VOCAB)]] for i in range(n))
+
+
+def calibrate():
+    """Measure words->tokens from the server's own counter."""
+    return 2000.0 / count_tokens(words(2000))
+
+
+def prompt_for(target_tokens, wpt):
+    n = max(1, int(target_tokens * wpt))
+    text = words(n)
+    pt = count_tokens(text)
+    while pt > PROMPT_CEILING:
+        n = max(1, int(n * (PROMPT_CEILING / pt) * 0.98))
+        text = words(n)
+        pt = count_tokens(text)
+    return text, pt
+
+
+def measured(prompt, max_tokens, temperature=0):
+    """Return (ttft_s, decode_tok_s, completion_tokens) via usage counters."""
+    body = {"model": MODEL,
+            "messages": [{"role": "user", "content": prompt}],
+            "max_tokens": max_tokens, "temperature": temperature,
+            "stream": True, "stream_options": {"include_usage": True}}
+    req = urllib.request.Request(
+        BASE + "/v1/chat/completions", data=json.dumps(body).encode(),
+        headers={"Content-Type": "application/json"})
+    t0 = time.time()
+    ttft = None
+    usage = {}
+    with urllib.request.urlopen(req, timeout=1800) as r:
+        for raw in r:
+            line = raw.decode("utf-8", "replace").strip()
+            if not line.startswith("data:"):
+                continue
+            s = line[5:].strip()
+            if s == "[DONE]":
+                break
+            try:
+                ev = json.loads(s)
+            except json.JSONDecodeError:
+                continue
+            if ev.get("usage"):
+                usage = ev["usage"]
+            for ch in ev.get("choices", []):
+                if (ch.get("delta") or {}).get("content"):
+                    if ttft is None:
+                        ttft = time.time() - t0
+    wall = time.time() - t0
+    ct = int(usage.get("completion_tokens") or 0)
+    if ct <= 0:
+        # No usage on the stream: the endpoint is unhealthy or the model
+        # returned nothing (e.g. HTTP 200 with an error payload).
+        raise RuntimeError("no completion_tokens in usage: %r" % (usage,))
+    dec_window = max(wall - (ttft or 0.0), 1e-9)
+    return (ttft or wall), (ct / dec_window), ct
+
+
+def fmt_temp(t):
+    return ("%.1f" % t).rstrip("0").rstrip(".") if t else "0"
+
+
+def main():
+    log("base=%s model=%s ceiling=%d temps=%s"
+        % (BASE, MODEL, PROMPT_CEILING, ",".join(fmt_temp(t) for t in TEMPS)))
+    try:
+        wpt = calibrate()
+    except Exception as e:
+        log("endpoint unhealthy during calibration: %s" % e)
+        return 2
+    log("words/token: %.3f" % wpt)
+
+    result = {"words_per_token": round(wpt, 5)}
+    for tgt in DEPTHS:
+        try:
+            p, pt = prompt_for(tgt, wpt)
+            measured(p, WARM_TOKENS)  # warm the arena and the prefix cache
+        except Exception as e:
+            log("warm pass failed at depth %d: %s" % (tgt, e))
+            return 2
+        for temp in TEMPS:
+            suffix = "" if temp == 0 else "_t%s" % fmt_temp(temp)
+            rates, tts, cts = [], [], []
+            for i in range(RUNS):
+                try:
+                    tt, r, ct = measured(p, MAX_TOKENS, temp)
+                except Exception as e:
+                    log("run %d/%d failed at depth %d temp %s: %s"
+                        % (i + 1, RUNS, tgt, fmt_temp(temp), e))
+                    continue
+                rates.append(r)
+                tts.append(tt)
+                cts.append(ct)
+            if len(rates) < 2:
+                log("only %d usable run(s) at depth %d temp %s — refusing to median"
+                    % (len(rates), tgt, fmt_temp(temp)))
+                return 2
+            result["decode_tps_%d%s" % (tgt, suffix)] = round(statistics.median(rates), 2)
+            result["ttft_ms_%d%s" % (tgt, suffix)] = round(1000.0 * statistics.median(tts), 1)
+            result["completion_tokens_%d%s" % (tgt, suffix)] = int(statistics.median(cts))
+            result["prompt_tokens_%d%s" % (tgt, suffix)] = pt
+            log("depth %5d tok temp %-4s decode %7.2f t/s  ttft %.3f s  (runs=%d)"
+                % (pt, fmt_temp(temp), result["decode_tps_%d%s" % (tgt, suffix)],
+                   result["ttft_ms_%d%s" % (tgt, suffix)] / 1000.0, len(rates)))
+
+    for temp in TEMPS:
+        suffix = "" if temp == 0 else "_t%s" % fmt_temp(temp)
+        meds = [result["decode_tps_%d%s" % (t, suffix)] for t in DEPTHS]
+        result["decode_tps%s" % suffix] = round(statistics.median(meds), 2)
+    print(json.dumps(result))
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

scripts/autoresearch/frank-primary-window.sh

@@ -0,0 +1,218 @@
+#!/usr/bin/env bash
+# frank-primary-window.sh — one full bench window for frank-primary, run ON
+# DANKSERVER as the deploy user (invoked over ssh by ./autoresearch.sh).
+#
+# Discipline (AGENTS.md Benchmarking Policy): the production unit is
+# STOPPED + MASKED first, the candidate config boots AS THE SERVICE itself
+# (same serve script, same container name, same port 7770 — no scratch
+# ports, no co-tenant GPU), the in-container measure runs, and the window
+# ALWAYS ends by restoring the production unit and verifying its health.
+# The EXIT trap makes the restore unconditional: a boot failure, a measure
+# failure, or a dead ssh session all end with production back up.
+#
+# Staging layout (created by the orchestrator, ephemeral — never a config
+# source; the repo checkout is):
+#   /home/dank/.autoresearch-frank-primary/
+#     serve.sh             staged candidate serve script
+#     frank-primary.env    staged candidate env (the candidate knobs)
+#     measure.py           the in-container measure client
+#     hash                 candidate fingerprint
+#     run.json             result (metrics + boot + kv + fingerprint)
+#     serve.out/measure.out/fail.log
+#     last/                memo: run.json + hash of the last FULLY
+#                          successful window (bench AND verified restore)
+#
+# This script logs to stderr and exits 0 only when the measure produced a
+# usable result (the restore happens in the trap regardless of the bench
+# outcome; a failed restore forces exit 1).
+set -uo pipefail
+
+STAGE="${FRANK_PRIMARY_AR_STAGE:-/home/dank/.autoresearch-frank-primary}"
+UNIT="${FRANK_PRIMARY_AR_UNIT:-frank-primary}"
+CTR="${FRANK_PRIMARY_AR_CTR:-frank-primary-qwen}"
+HEALTH_URL="http://127.0.0.1:7770/health"
+MODELS_URL="http://127.0.0.1:7770/v1/models"
+BOOT_TIMEOUT="${FRANK_PRIMARY_AR_BOOT_TIMEOUT:-1200}"
+RESTORE_TIMEOUT="${FRANK_PRIMARY_AR_RESTORE_TIMEOUT:-1200}"
+MEASURE_TIMEOUT="${FRANK_PRIMARY_AR_MEASURE_TIMEOUT:-1800}"
+SERVED="$(sed -n 's/^FRANK_PRIMARY_SERVED_NAME=//p' "$STAGE/frank-primary.env" 2>/dev/null | head -1)"
+SERVED="${SERVED:-frank-primary}"
+
+RESTORED=0
+
+log() { printf '[window %s] %s\n' "$(date '+%H:%M:%S')" "$*" >&2; }
+
+container_state() { podman inspect -f '{{.State.Status}}' "$CTR" 2>/dev/null || echo absent; }
+
+wait_port_free() {
+    local i
+    for i in $(seq 1 90); do
+        [ "$(container_state)" != "running" ] && return 0
+        sleep 2
+    done
+    return 1
+}
+
+wait_health() { # $1 = timeout seconds
+    local t0 code
+    t0=$(date +%s)
+    while :; do
+        code="$(curl -s -o /dev/null -w '%{http_code}' --max-time 10 "$HEALTH_URL" 2>/dev/null || echo 000)"
+        [ "$code" = "200" ] && return 0
+        [ "$(( $(date +%s) - t0 ))" -ge "$1" ] && return 1
+        sleep 2
+    done
+}
+
+capture_bench_failure() {
+    log "capturing bench-boot failure evidence"
+    {
+        echo "=== podman logs (tail 400) ==="
+        podman logs --tail 400 "$CTR" 2>&1 || true
+        echo "=== staged serve.out (tail 60) ==="
+        tail -60 "$STAGE/serve.out" 2>/dev/null || true
+        echo "=== engine log (tail 60) ==="
+        tail -60 /home/dank/logs/frank_primary.log 2>/dev/null || true
+    } >"$STAGE/fail.log" 2>&1 || true
+    grep -E 'OutOfMemory|Tried to allocate|Engine core initialization failed|Error|error' \
+        "$STAGE/fail.log" 2>/dev/null | tail -25 >&2 || true
+}
+
+restore() {
+    [ "$RESTORED" = 1 ] && return 0
+    RESTORED=1
+    log "restoring production unit $UNIT"
+    if [ -f "$STAGE/serve.pid" ]; then
+        kill "$(cat "$STAGE/serve.pid" 2>/dev/null)" 2>/dev/null || true
+    fi
+    podman stop -t 120 "$CTR" >/dev/null 2>&1 || true
+    wait_port_free || log "WARNING: bench container still draining after 180s"
+    sudo systemctl unmask "$UNIT" 2>/dev/null || true
+    if ! sudo systemctl start "$UNIT"; then
+        log "ERROR: systemctl start $UNIT failed — operator intervention required"
+        return 1
+    fi
+    if wait_health "$RESTORE_TIMEOUT"; then
+        if curl -s --max-time 10 "$MODELS_URL" 2>/dev/null | grep -q "$SERVED"; then
+            log "production restored and verified ($SERVED answering on :7770)"
+            # Memo promotion lives HERE: a cached result is only ever the
+            # product of a fully successful window (bench AND verified
+            # restore), never of a bare bench success.
+            cp "$STAGE/run.json" "$STAGE/last/" 2>/dev/null || true
+            cp "$STAGE/hash" "$STAGE/last/" 2>/dev/null || true
+        else
+            log "ERROR: health 200 but /v1/models does not list $SERVED"
+            return 1
+        fi
+    else
+        log "ERROR: production not healthy within ${RESTORE_TIMEOUT}s — unit left to systemd (Restart=always)"
+        return 1
+    fi
+    return 0
+}
+trap 'restore; rc=$?; [ "$rc" -ne 0 ] && exit 1' EXIT
+
+main() {
+    local f
+    for f in serve.sh frank-primary.env measure.py hash; do
+        [ -f "$STAGE/$f" ] || { log "staged file missing: $STAGE/$f"; exit 1; }
+    done
+    mkdir -p "$STAGE/last"
+    rm -f "$STAGE/run.json" "$STAGE/fail.log" "$STAGE/measure.out" "$STAGE/evidence.log"
+
+    local t_start t0 i kvline kv_tokens kv_conc measure_json boot_seconds
+    t_start="$(date +%s)"
+
+    log "stopping and masking production unit $UNIT"
+    sudo systemctl stop "$UNIT" 2>/dev/null || true
+    # Explicit stop is not auto-restarted even with Restart=always; the mask
+    # is belt-and-braces against a stray start during the window.
+    sudo systemctl mask "$UNIT" 2>/dev/null || \
+        log "WARNING: mask failed — relying on explicit-stop semantics"
+    wait_port_free || log "note: production container still draining"
+
+    t0="$(date +%s)"
+    log "starting bench instance from staged config (fingerprint $(tr -d '[:space:]' <"$STAGE/hash"))"
+    FRANK_PRIMARY_ENV_FILE="$STAGE/frank-primary.env" \
+        setsid nohup bash "$STAGE/serve.sh" >"$STAGE/serve.out" 2>&1 </dev/null &
+    echo "$!" >"$STAGE/serve.pid"
+
+    for i in $(seq 1 45); do
+        [ "$(container_state)" = "running" ] && break
+        sleep 2
+    done
+    if [ "$(container_state)" != "running" ]; then
+        log "bench container never reached running state"
+        tail -40 "$STAGE/serve.out" >&2 || true
+        exit 1
+    fi
+
+    if wait_health "$BOOT_TIMEOUT"; then
+        boot_seconds=$(( $(date +%s) - t0 ))
+        log "bench healthy after ${boot_seconds}s"
+    else
+        if [ "$(container_state)" = "running" ]; then
+            log "bench boot timed out after ${BOOT_TIMEOUT}s (still loading)"
+        else
+            log "bench container died during boot"
+            capture_bench_failure
+        fi
+        exit 1
+    fi
+
+    log "running measure in container (timeout ${MEASURE_TIMEOUT}s)"
+    if ! podman cp "$STAGE/measure.py" "$CTR":/tmp/ar-measure.py; then
+        log "podman cp of measure client failed"
+        exit 1
+    fi
+    if ! timeout "$MEASURE_TIMEOUT" podman exec "$CTR" python3 /tmp/ar-measure.py \
+            >"$STAGE/measure.out" 2>&1; then
+        log "measure failed or timed out"
+        tail -40 "$STAGE/measure.out" >&2 || true
+        exit 1
+    fi
+
+    # Capacity facts for the run (context-size experiments change these).
+    kvline="$(podman logs "$CTR" 2>&1 | grep -E 'GPU KV cache size' | tail -1 || true)"
+    kv_tokens="$(printf '%s' "$kvline" | sed -E 's/.*GPU KV cache size: ([0-9,]+) tokens.*/\1/' | tr -d ',' || true)"
+    kv_conc="$(printf '%s' "$kvline" | sed -E 's/.*per request: ([0-9.]+x).*/\1/' || true)"
+
+    # Human-facing evidence, in a sidecar file because it is multi-line prose,
+    # not metrics: the expert-plan lines (the ONLY place the arena / runtime
+    # overhead / KV-reserve numbers exist — they are never in /metrics), the
+    # compile path (cold build vs cache hit), the graph-capture sizes, the
+    # speculative-decode counters, and any OOM signature. Without this an arm
+    # that loses decode speed is unexplainable afterwards.
+    {
+        echo "=== engine plan / capacity / spec evidence ==="
+        podman logs "$CTR" 2>&1 | grep -E \
+'KV cache size|Maximum concurrency|layers split|expert cache|runtime overhead|non-default args|r4d all-reduce|Using \[|cudagraph|CUDA graph|Capturing|graph for compile range|torch.compile took|Directly load the compiled|speculative|Speculative|acceptance|Acceptance|offload|precise|OutOfMemory|Tried to allocate|Engine core initialization failed' \
+            | tail -150 || true
+        echo "=== /metrics counters (spec decode, caches, success) ==="
+        curl -s --max-time 15 http://127.0.0.1:7770/metrics 2>/dev/null \
+            | grep -E '^vllm:(spec_decode|gpu_cache|prefix_cache|request_success|generation_tokens|iteration)' \
+            | head -80 || true
+    } >"$STAGE/evidence.log" 2>&1 || true
+    log "engine evidence -> $STAGE/evidence.log ($(wc -l <"$STAGE/evidence.log") lines)"
+
+    measure_json="$(grep -E '^\{' "$STAGE/measure.out" | tail -1 || true)"
+    if [ -z "$measure_json" ]; then
+        log "measure produced no JSON result line"
+        tail -40 "$STAGE/measure.out" >&2 || true
+        exit 1
+    fi
+
+    # The measure line is already JSON, so the run object can be assembled
+    # without any host-side interpreter.
+    printf '{"hash":"%s","boot_seconds":%s,"kv_tokens":%s,"kv_concurrency":"%s","total_seconds":%s,"measure":%s}\n' \
+        "$(tr -d '[:space:]' <"$STAGE/hash")" \
+        "$boot_seconds" "${kv_tokens:-0}" "${kv_conc:-unknown}" \
+        "$(( $(date +%s) - t_start ))" "$measure_json" >"$STAGE/run.json"
+
+    log "bench window complete in $(( $(date +%s) - t_start ))s"
+    # Memo only after a FULL success; the restore below still has to verify.
+    exit 0
+}
+
+main "$@"
+exit "$?"

scripts/build-gufo.sh

@@ -0,0 +1,96 @@
+#!/usr/bin/env bash
+# Build the gufo runtime that serves frank-large on Dankclaw.
+#
+#   bash scripts/build-gufo.sh              # build + install the pinned tag
+#   bash scripts/build-gufo.sh --verify     # only check the installed tree
+#
+# gufo (gufo-org/gufo) is the Strix Halo inference engine: it targets
+# gfx1151 specifically and has no prebuilt Linux release binaries, so the
+# runtime is built from the pinned git tag and installed to
+# /home/dank/GitPackages/gufo-<tag>/. Everything below is what Dankclaw
+# needs on top of AMD's Linux instructions.
+#
+# The host already carries the ROCm 7.2.4 stack (/opt/rocm -> /etc/
+# alternatives/rocm) with the HIP dev packages, so this script installs
+# only the ordinary system libraries gufo links against.
+#
+# Two host facts this build depends on, both discovered on 2026-09-28:
+#
+#   1. Ubuntu's librocwmma-dev ships a TRUNCATED rocWMMA header set (the
+#      public headers without rocwmma/internal/) into /usr/include, where
+#      it shadows AMD's complete copy under /opt/rocm/include. The first
+#      build died on "'internal/accessors.hpp' file not found" from
+#      /usr/include/rocwmma/rocwmma.hpp. A -I cannot fix it: in HIP mode
+#      the host compilation pass searches the system directories first.
+#      The package has no reverse dependencies, so it is removed.
+#   2. The ROCm clang picks the NEWEST /usr/include/c++ tree, which on
+#      Ubuntu 26.04 is GCC 16's. Its <atomic_wait.h> emits a symbol
+#      reference the installed libstdc++ (GLIBCXX_3.4.35, GCC 15) does not
+#      export, and the link fails with "undefined symbol:
+#      std::__detail::__notify_impl". --gcc-install-dir pins the headers to
+#      the same GCC the runtime library came from.
+set -euo pipefail
+
+GUFO_TAG="${GUFO_TAG:-v0.1.1}"
+GUFO_REPO="${GUFO_REPO:-https://github.com/gufo-org/gufo}"
+# The commit the tag resolved to when this runtime was qualified. A tag that
+# moves is an unreproducible runtime, so the pin is checked, not assumed.
+GUFO_COMMIT="${GUFO_COMMIT:-b0f8673470a3dcc2b5e9c9546c48a40b82b5ff31}"
+SRC_DIR="${GUFO_SRC_DIR:-/home/dank/GitPackages/gufo-src}"
+PREFIX="${GUFO_PREFIX:-/home/dank/GitPackages/gufo-${GUFO_TAG}}"
+JOBS="${JOBS:-$(nproc)}"
+# The GCC toolchain that owns the installed libstdc++ on this host.
+GCC_DIR="${GCC_DIR:-/usr/lib/gcc/x86_64-linux-gnu/15}"
+
+# check_pin — refuse a source tree that is not the qualified commit.
+check_pin() {
+    local commit
+    commit="$(git -C "$SRC_DIR" rev-parse HEAD 2>/dev/null || true)"
+    if [[ -n "$commit" && "$commit" != "$GUFO_COMMIT" ]]; then
+        echo "GUFO_COMMIT mismatch: $SRC_DIR is at $commit, pinned $GUFO_COMMIT" >&2
+        exit 1
+    fi
+}
+
+if [[ "${1:-}" == "--verify" ]]; then
+    # The pin first: a moved tag is the more urgent thing to hear about, and
+    # it is the check that must not be masked by a missing install.
+    check_pin
+    [[ -x "$PREFIX/bin/gufo" ]] || { echo "no gufo at $PREFIX/bin/gufo" >&2; exit 1; }
+    "$PREFIX/bin/gufo" diagnose | tail -n 5
+    exit 0
+fi
+
+echo "== system libraries gufo links against =="
+sudo -n apt-get install -y -qq --no-install-recommends \
+    build-essential cmake ninja-build pkg-config \
+    libicu-dev libcurl4-openssl-dev libssl-dev libpng-dev libjpeg-dev ffmpeg
+
+echo "== dropping Ubuntu's truncated rocWMMA headers =="
+if dpkg -s librocwmma-dev >/dev/null 2>&1; then
+    sudo -n apt-get remove -y -qq librocwmma-dev
+fi
+
+echo "== source: $GUFO_REPO @ $GUFO_TAG =="
+if [[ ! -d "$SRC_DIR/.git" ]]; then
+    mkdir -p "$(dirname "$SRC_DIR")"
+    git clone --depth 1 --branch "$GUFO_TAG" "$GUFO_REPO" "$SRC_DIR"
+fi
+git -C "$SRC_DIR" fetch --depth 1 origin "refs/tags/$GUFO_TAG"
+git -C "$SRC_DIR" checkout -q --detach "refs/tags/$GUFO_TAG"
+echo "   $(git -C "$SRC_DIR" rev-parse HEAD)"
+check_pin
+
+echo "== configure + build (release, HIP gfx1151) =="
+cd "$SRC_DIR"
+toolchain="--gcc-install-dir=$GCC_DIR"
+cmake --preset release \
+    -DCMAKE_INSTALL_PREFIX="$PREFIX" \
+    -DCMAKE_HIP_FLAGS="$toolchain" \
+    -DCMAKE_EXE_LINKER_FLAGS="$toolchain"
+cmake --build --preset release --parallel "$JOBS"
+cmake --install build/release
+
+echo "== verify =="
+"$PREFIX/bin/gufo" diagnose
+echo "frank-large runtime ready: $PREFIX/bin/gufo ($GUFO_TAG @ $commit)"

scripts/build-rocmfpx.sh

@@ -0,0 +1,89 @@
+#!/usr/bin/env bash
+# Build the ROCmFPX runtime that serves frank-large on Dankclaw.
+#
+#   bash scripts/build-rocmfpx.sh              # clone/fetch + build the pin
+#   bash scripts/build-rocmfpx.sh --verify     # only check the installed tree
+#
+# Runtime: kingjones30/ROCmFPX — charlie12345/ROCmFPX plus the model-card
+# patches, i.e. the fork that owns the ROCmFPx tensor formats. frank-large's
+# payload is ftype 115 (`Q8_0_ROCMFPX_AGENT`): 311 tensors of
+# GGML_TYPE_Q8_0_ROCMFPX (103), a type that exists ONLY in this fork, so
+# stock llama.cpp, gufo and every other loader on the host reject the file.
+# This is the same tree that served the Qwen3.8-Flash-Next ROCmFP4 STRIX
+# endpoint 2026-08-30 → 09-11 (see the kingjones runbook), rebuilt here so a
+# fresh host can be brought up from the repo alone.
+#
+# Why this commit and not a newer one: 36e9acd (2026-08-28) is what was
+# qualified against this payload on 2026-10-02 — it loads ftype 115, survives
+# warmup, and drafts with the MTP head at acceptance 0.65. The five legacy
+# builds of 2026-08-25 crashed this same file in ggml_compute_forward_set_rows
+# (see kingjones-q8agent-rocmfpx.md), so the pin is load-bearing: do not bump
+# it without re-running that crash test.
+#
+# Host notes:
+#   * -DLLAMA_BUILD_WEBUI=OFF is REQUIRED on this box. The tree provisions its
+#     embedded UI assets via npm or the HF llama-ui bucket; npm is absent here
+#     and the bucket download half-failed, aborting the link at 98%. The API
+#     server is unaffected.
+#   * The build links against the host ROCm (/opt/rocm); no bundled toolchain
+#     like the gfx1201 R9700 build needed.
+set -euo pipefail
+
+ROCMFPX_REPO="${ROCMFPX_REPO:-https://github.com/kingjones30/ROCmFPX.git}"
+# The commit frank-large was qualified with (measured 2026-10-02).
+ROCMFPX_COMMIT="${ROCMFPX_COMMIT:-36e9acd40e10a87cd3c3ef8ec734668757dc8520}"
+SRC_DIR="${ROCMFPX_SRC_DIR:-/home/dank/GitPackages/kingjones-ROCmFPX}"
+BUILD_DIR="${ROCMFPX_BUILD_DIR:-$SRC_DIR/build}"
+GPU_TARGET="${ROCMFPX_GPU_TARGET:-gfx1151}"
+JOBS="${JOBS:-$(nproc)}"
+
+# check_pin — refuse a source tree that is not the qualified commit. A moved
+# fork means different kernels under the same path, and on this model line a
+# wrong build is a segfault at first request, not a clean error.
+check_pin() {
+    local commit
+    commit="$(git -C "$SRC_DIR" rev-parse HEAD 2>/dev/null || true)"
+    if [[ -n "$commit" && "$commit" != "$ROCMFPX_COMMIT" ]]; then
+        echo "ROCMFPX_COMMIT mismatch: $SRC_DIR is at $commit, pinned $ROCMFPX_COMMIT" >&2
+        exit 1
+    fi
+}
+
+if [[ "${1:-}" == "--verify" ]]; then
+    check_pin
+    [[ -x "$BUILD_DIR/bin/llama-server" ]] || {
+        echo "no ROCmFPX llama-server at $BUILD_DIR/bin/llama-server" >&2
+        exit 1
+    }
+    # The format the endpoint exists for: the fork must carry the ROCmFPx
+    # tensor type, or the payload cannot load at all.
+    "$BUILD_DIR/bin/llama-server" --help 2>&1 | grep -q -- '--spec-type' || {
+        echo "$BUILD_DIR/bin/llama-server has no --spec-type: MTP drafting is impossible" >&2
+        exit 1
+    }
+    "$BUILD_DIR/bin/llama-server" --version | head -1
+    exit 0
+fi
+
+echo "== source: $ROCMFPX_REPO @ $ROCMFPX_COMMIT =="
+if [[ ! -d "$SRC_DIR/.git" ]]; then
+    mkdir -p "$(dirname "$SRC_DIR")"
+    git clone "$ROCMFPX_REPO" "$SRC_DIR"
+fi
+git -C "$SRC_DIR" fetch --quiet origin "$ROCMFPX_COMMIT" || true
+git -C "$SRC_DIR" checkout -q --detach "$ROCMFPX_COMMIT"
+check_pin
+echo "   $(git -C "$SRC_DIR" rev-parse HEAD)"
+
+echo "== configure + build (release, HIP $GPU_TARGET) =="
+cmake -S "$SRC_DIR" -B "$BUILD_DIR" \
+    -DGGML_HIP=ON \
+    -DGPU_TARGETS="$GPU_TARGET" \
+    -DGGML_NATIVE=ON \
+    -DCMAKE_BUILD_TYPE=Release \
+    -DLLAMA_BUILD_WEBUI=OFF
+cmake --build "$BUILD_DIR" --target llama-server llama-quantize -j "$JOBS"
+
+echo "== verify =="
+"$BUILD_DIR/bin/llama-server" --version | head -1
+echo "frank-large runtime ready: $BUILD_DIR/bin/llama-server ($ROCMFPX_COMMIT)"

scripts/dankstatus.sh

@@ -0,0 +1,71 @@
+#!/usr/bin/env bash
+# Single source of truth for Dankserver GLM-5.3 state.
+#
+# This exists because the ad-hoc status commands kept producing WRONG readings.
+# The specific trap: /proc/meminfo separates key from value with a TAB, so the
+# common idiom
+#   grep MemAvailable /proc/meminfo | tr -s " " | cut -d" " -f2
+# yields an empty field (tab is not a space) and awk then prints 0.0 -- which
+# reads as "host out of memory" when the host actually has 240 GiB free.
+# Everything below parses on a colon split, which is tab-agnostic, and the
+# memory figure is cross-checked before it is shown.
+set -uo pipefail
+
+MB=1048576
+GLM_CTR="${GLM_CTR:-glm-w4a16}"
+
+mem_field() { awk -v k="$1:" '$1==k {printf "%.1f", $2/'"$MB"'; exit}' /proc/meminfo; }
+section() { printf '\n== %s ==\n' "$1"; }
+
+section "time / host"
+printf '  %s   %s\n' "$(date +%H:%M:%S)" \
+  "$(awk -F: '/model name/{print $2; exit}' /proc/cpuinfo | sed 's/^ *//')"
+
+section "memory (GiB)"
+TOT=$(mem_field MemTotal); AVAIL=$(mem_field MemAvailable)
+printf '  MemTotal     %8s\n' "$TOT"
+printf '  MemAvailable %8s\n' "$AVAIL"
+printf '  Cached       %8s\n' "$(mem_field Cached)"
+awk -v t="$TOT" -v a="$AVAIL" 'BEGIN{
+  if (t+0 <= 0 || a+0 <= 0) { print "  *** PARSE BUG: zero value ***"; exit }
+  if (a+0 > t+0)         { print "  *** PARSE BUG: available > total ***"; exit }
+  printf "  check        ok (%.0f%% free)\n", 100*a/t
+}'
+
+section "swap"
+swapon --show=NAME,SIZE,USED --noheadings 2>/dev/null \
+  | awk '{printf "  %-12s %6s used of %s\n",$1,$3,$2}' || echo "  none"
+
+section "gpu"
+# CSV layout: device,VRAM Total Memory (B),VRAM Total Used Memory (B)
+rocm-smi --showmeminfo vram --csv 2>/dev/null | awk -F, '
+  NR>1 && $1 ~ /^card[0-9]+$/ {
+    printf "  %-8s %7.2f GiB used of %.2f\n", $1, $3/1073741824, $2/1073741824
+  }'
+printf '  compute util: '
+rocm-smi --showuse --csv 2>/dev/null | awk -F, 'NR>1 && $1 ~ /^card[0-9]+$/ {printf "%s=%s%% ", $1, $2}'
+echo
+
+section "disk"
+df -h /home/dank 2>/dev/null | awk 'NR==2{printf "  %s  %s used  %s free  (%s)\n",$1,$3,$4,$5}'
+
+section "llm containers"
+if podman ps -a --format '{{.Names}}' 2>/dev/null | grep -qE 'glm|frank'; then
+  podman ps -a --format '{{.Names}}|{{.Status}}' 2>/dev/null \
+    | awk -F'|' '$1 ~ /glm|frank/ {printf "  %-18s %s\n",$1,$2}'
+else
+  echo "  none"
+fi
+
+section "frank-primary unit"
+printf '  active:  %s\n' "$(systemctl is-active frank-primary 2>/dev/null | head -1)"
+printf '  enabled: %s\n' "$(systemctl is-enabled frank-primary 2>/dev/null | head -1)"
+
+section "glm serve tail"
+if podman ps -a --format '{{.Names}}' 2>/dev/null | grep -qx "$GLM_CTR"; then
+  podman logs "$GLM_CTR" 2>&1 | grep -viE 'multiproc_executor\.py:943' | tail -5 | sed 's/^/  /'
+else
+  echo "  (container $GLM_CTR not present)"
+fi
+
+section "done"

scripts/deploy-remote.sh

@@ -935,15 +935,24 @@ deploy_plex() {
     fi
     systemctl --user restart plex.service
     # Readiness gate: /identity answers MediaContainer XML once PMS is up.
+    #
+    # HTTPS, NOT HTTP. PMS drops plain HTTP off-loopback: the probe's
+    # connection arrives on the container's eth0, so an http:// request is
+    # accepted and then answered with nothing (curl 52, Empty reply) against a
+    # perfectly healthy server. Measured 2026-10-02 — the gate burned its full
+    # 600s and failed a plex whose https://127.0.0.1:32400/identity answered
+    # 200 in 44 ms the whole time. The same law is why the Caddyfile dials the
+    # plex upstream over TLS (tests/caddyfile.bats pins it).
+    # -k: PMS presents its own self-signed cert on 32400.
+    #
     # Reality check from the 2026-09 cutover: the FIRST container boot of a
-    # years-old library ran ~10 min (PMS pre-binds 32400 during boot and
-    # empty-replies until the HTTP layer is ready), so the gate allows 600s.
+    # years-old library ran ~10 min, so the gate still allows 600s.
     # Fail the deploy (pending state preserved) rather than mark it deployed blind.
     for ((i = 1; i <= 120; i++)); do
-        curl -fsS --max-time 3 http://127.0.0.1:32400/identity >/dev/null && return 0
+        curl -fsSk --max-time 3 https://127.0.0.1:32400/identity >/dev/null && return 0
         sleep 5
     done
-    echo "FAIL: plex did not answer /identity within 600s" >&2
+    echo "FAIL: plex did not answer /identity over HTTPS within 600s" >&2
     return 1
 }
 
@@ -1077,20 +1086,33 @@ deploy_anubis() {
     wait_container_healthy anubis 60 || return 1
 }
 
-# frank-large is the stock llama.cpp prebuilt (unsloth b11030, ROCm
-# gfx1151) serving unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL with the DFlash2
-# drafter (z-lab/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M) on the Strix Halo iGPU,
-# as a ROOT unit since the 2026-09-19 cutover: the runtime is a prebuilt
-# binary, so there is no container image to install and no generated unit
-# to dodge. relink_unit points /etc/systemd/system/frank-large.service at
-# the repo file and boot persistence is that unit's [Install] section.
+# frank-large is the halogen CONTAINER (ghcr.io/peonist-ai/halogen,
+# digest-pinned through scripts/frank-large.env) serving
+# peonist-ai/halogen-qwen3.8-27b — qwen3.8-27b-p1w4d-d2.hgn, 33.4 GiB,
+# DFlash2 spec decode, 1 slot x 200700, text-only — on the Strix Halo
+# iGPU. It stays a ROOT unit because the serve script drives `podman run`
+# directly (the frank-primary pattern): relink_unit points
+# /etc/systemd/system/frank-large.service at the repo file, boot
+# persistence is that unit's [Install], and LimitMEMLOCK=infinity is
+# fleet parity. 2026-10-05 cutover from the ROCmFPX llama.cpp shape —
+# rollback = revert the swap commit's trio; that payload, its draft head
+# and the fork build all stay on disk
+# (~/models/kingjones-Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP,
+# ~/GitPackages/kingjones-ROCmFPX).
 #
-# It replaces the Kairic Edge runtime (ciru-ai/ROCmFPX), which had
-# replaced the unsloth llama.cpp prebuilt serving
-# unsloth/Qwen3.8-Flash-Next-GGUF:UD-IQ4_XS. The port gate is the same as
-# every cutover since 2026-09-16: nothing starts until 7770 is released.
-LLM_LARGE_MODEL="${LLM_LARGE_MODEL:-/home/dank/models/Qwen3.8-27B-UD-Q8_K_XL/Qwen3.8-27B-UD-Q8_K_XL.gguf}"
-LLM_LARGE_DRAFT="${LLM_LARGE_DRAFT:-/home/dank/models/Qwen3.8-27B-UD-Q8_K_XL/Qwen3.8-27B-DFlash2-Q4_K_M.gguf}"
+# The gates mirror serve-frank-large.sh's predicates EXACTLY (the
+# 2026-10-01 crash-loop lesson): image digest in dank's podman storage,
+# the .hgn checkpoint at the exact byte count, the tokenizer directory
+# holding every named file as a regular (not symlinked) file — a green
+# deploy against a missing artifact would
+# leave a restart-looping unit and a dark endpoint. The port gate is the
+# same as every cutover since 2026-09-16: nothing starts until 7770 is
+# released.
+LLM_LARGE_IMAGE="${LLM_LARGE_IMAGE:-ghcr.io/peonist-ai/halogen@sha256:dc0a39a0016d6cfc58a197978febaafdf8d28403f724ded6111d98b5fb7ac0ea}"
+LLM_LARGE_CKPT="${LLM_LARGE_CKPT:-/home/dank/models/halogen-qwen3.8-27b/qwen3.8-27b-p1w4d-d2.hgn}"
+LLM_LARGE_CKPT_BYTES="${LLM_LARGE_CKPT_BYTES:-35865565184}"
+LLM_LARGE_TOK_DIR="${LLM_LARGE_TOK_DIR:-/home/dank/models/halogen-qwen3.8-27b/tokenizer}"
+LLM_LARGE_TOK_FILES="${LLM_LARGE_TOK_FILES:-tokenizer.json chat_template.jinja tokenizer_config.json merges.txt vocab.json}"
 
 retire_halogen_large_quadlet() {
     local trash="$HOME/.AGENTTRASH/20260919-orca-cutover/quadlet"
@@ -1123,34 +1145,51 @@ wait_port_free() {
 }
 
 deploy_llm_large() {
-    # Refuse rather than restart-loop. A unit whose runtime or payload is
-    # missing never becomes healthy, and a green deploy would leave the
-    # endpoint dark with nothing in the deploy log to say why.
-    local missing=0 f
-    for f in "$LLM_LARGE_MODEL" "$LLM_LARGE_DRAFT"; do
-        if [[ ! -f "$f" ]]; then
-            echo "ERROR: payload absent at $f" >&2
-            echo "frank-large: fetch it: bash scripts/download-frank-large.sh" >&2
-            missing=1
+    # Refuse rather than restart-loop; every gate below has a twin in
+    # serve-frank-large.sh and they must agree (2026-10-01 lesson).
+    # Capture podman output before grepping: `podman images | grep -q`
+    # races under pipefail (SIGPIPE 141 becomes "no match", 2026-09-30).
+    if ! grep -qF "${LLM_LARGE_IMAGE#*@}" <<<"$(podman images --format '{{.RepoDigests}}' 2>/dev/null)"; then
+        echo "ERROR: halogen image absent: $LLM_LARGE_IMAGE" >&2
+        echo "frank-large: pull it: podman pull ${LLM_LARGE_IMAGE%@*}" >&2
+        return 1
+    fi
+    if [[ ! -f "$LLM_LARGE_CKPT" ]]; then
+        echo "ERROR: halogen checkpoint absent at $LLM_LARGE_CKPT" >&2
+        echo "frank-large: fetch it: bash scripts/download-frank-large-halogen.sh" >&2
+        return 1
+    fi
+    if [[ "$(stat -c %s "$LLM_LARGE_CKPT")" != "$LLM_LARGE_CKPT_BYTES" ]]; then
+        echo "ERROR: halogen checkpoint is not $LLM_LARGE_CKPT_BYTES bytes (partial download?)" >&2
+        echo "frank-large: re-check: bash scripts/download-frank-large-halogen.sh --verify-only" >&2
+        return 1
+    fi
+    if [[ -L "$LLM_LARGE_TOK_DIR" || ! -d "$LLM_LARGE_TOK_DIR" ]]; then
+        echo "ERROR: tokenizer dir missing or a symlink at $LLM_LARGE_TOK_DIR" >&2
+        echo "frank-large: fetch it: bash scripts/download-frank-large-halogen.sh" >&2
+        return 1
+    fi
+    for tf in $LLM_LARGE_TOK_FILES; do
+        if [[ ! -f "$LLM_LARGE_TOK_DIR/$tf" || -L "$LLM_LARGE_TOK_DIR/$tf" ]]; then
+            echo "ERROR: tokenizer file missing or a symlink: $LLM_LARGE_TOK_DIR/$tf" >&2
+            echo "frank-large: fetch it: bash scripts/download-frank-large-halogen.sh" >&2
+            return 1
         fi
     done
-    [[ "$missing" -eq 0 ]] || return 1
 
     retire_halogen_large_quadlet
-    # On a re-deploy the predecessor is THIS unit: stop it before waiting on
-    # the port, or the wait deadlocks on the server it just stopped serving.
-    # Then sweep a stray manual run of the SAME model -- that is how a bench
-    # harness leaves the port busy. Match the model path, not the binary name:
-    # frank-small on this host is also a llama-server and must not be killed.
-    # The pattern never appears in the deploying shell's own command line, so
-    # -f is safe here (it is not in `bash deploy-remote.sh`).
+    # The predecessor on a re-deploy is THIS unit: stop it (ExecStop stops
+    # the container) before waiting on the port. The stray sweep is keyed
+    # on the FULL container name, which appears in the argv of a manual
+    # `podman run --name frank-large-halogen` / `podman wait` but NOT in
+    # this deployer's own command line (`bash deploy-remote.sh
+    # frank-large`) — so it cannot shoot itself in the foot (the fleet
+    # pkill law), and it can never match frank-small's llama-server.
     run_privileged systemctl stop frank-large.service 2>/dev/null || true
-    pkill -f 'Qwen3.8-27B-UD-Q8_K_XL' 2>/dev/null || true
-    # Nothing may start until the predecessor has released the port. On
-    # 2026-09-16 the cutover failed exactly here: the old server logged
-    # "Deactivated successfully", the next engine tried to bind 7770 a second
-    # later and reported "Address already in use". Gate on the port, not on
-    # hope.
+    pkill -f 'frank-large-halogen' 2>/dev/null || true
+    podman rm -f frank-large frank-large-halogen >/dev/null 2>&1 || true
+    # Nothing may start until the predecessor has released the port
+    # (2026-09-16: "Deactivated successfully" was not "port free").
     wait_port_free 7770 || return 1
     relink_unit frank-large
     run_privileged systemctl daemon-reload
@@ -1159,109 +1198,137 @@ deploy_llm_large() {
     wait_for_health frank-large http://127.0.0.1:7770/health 1200
 }
 
-# deploy_llm_small — frank-small is mainline llama.cpp on the RTX 3090, as a
-# ROOT unit: the runtime is a native CUDA binary, so there is no container
-# image to install and no generated unit to avoid. relink_unit points
-# /etc/systemd/system/frank-small.service at the repo file and boot persistence
-# is that unit's [Install] section.
+# deploy_llm_small — frank-small is llama.cpp (master build, CUDA sm_86) on
+# the RTX 3090 as a ROOT unit: the runtime is a native binary tree under
+# ~/GitPackages/llama-master/build-cuda, so there is no container image and
+# no generated unit to avoid. relink_unit points
+# /etc/systemd/system/frank-small.service at the repo file and boot
+# persistence is that unit's [Install] section.
 #
 # It is a SEPARATE endpoint on port 7771; frank-large keeps 7770. The 3090 is
 # not shared: frank-large's engine runs on the gfx1151 iGPU.
-LLM_SMALL_MODEL="${LLM_SMALL_MODEL:-/home/dank/models/Gemma4-12B-QAT/Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced-Q4_K_M.gguf}"
+LLM_SMALL_MODEL_FILE="${LLM_SMALL_MODEL_FILE:-/home/dank/models/Ornith-Q8/Ornith-1.5-9B-Q8_0.gguf}"
+LLM_SMALL_RUNTIME="${LLM_SMALL_RUNTIME:-/home/dank/GitPackages/llama-master/build-cuda/bin/llama-server}"
+# The projector IS the vision capability here: the Ornith GGUF ships no
+# `v.*` tensors, so a deploy that starts without this file silently ships a
+# text-only endpoint. Checked unconditionally, like frank-large's.
+LLM_SMALL_MMPROJ="${LLM_SMALL_MMPROJ:-/home/dank/models/Ornith-Q8/mmproj-Ornith-1.5-9B-BF16.gguf}"
 
 deploy_llm_small() {
-    if [[ ! -f "$LLM_SMALL_MODEL" ]]; then
-        echo "ERROR: model absent at $LLM_SMALL_MODEL" >&2
-        echo "frank-small: fetch HauhauCS/Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced (Q4_K_M, the mmproj and the MTP sidecar) first;" >&2
-        echo "frank-small: refusing to start a unit that cannot load." >&2
+    # Refuse rather than restart-loop: a unit whose runtime or payload is
+    # missing never becomes healthy, and a green deploy would leave the
+    # endpoint dark with nothing in the log to say why.
+    if [[ ! -x "$LLM_SMALL_RUNTIME" ]]; then
+        echo "ERROR: llama.cpp runtime absent at $LLM_SMALL_RUNTIME" >&2
+        echo "frank-small: build it (cmake -B build-cuda -DGGML_CUDA=ON" >&2
+        echo "frank-small: -DCMAKE_CUDA_ARCHITECTURES=86 -DCMAKE_CUDA_HOST_COMPILER=/usr/bin/gcc-13" >&2
+        echo "frank-small: && cmake --build build-cuda --target llama-server) first." >&2
         return 1
     fi
+    for f in "$LLM_SMALL_MODEL_FILE" "$LLM_SMALL_MMPROJ"; do
+        if [[ ! -f "$f" ]]; then
+            echo "ERROR: frank-small payload file absent at $f" >&2
+            echo "frank-small: run scripts/download-frank-small.sh first;" >&2
+            echo "frank-small: refusing to start a unit that cannot load." >&2
+            return 1
+        fi
+    done
     relink_unit frank-small
     run_privileged systemctl daemon-reload
     run_privileged systemctl enable frank-small.service
     run_privileged systemctl restart frank-small.service
-    wait_for_health frank-small http://127.0.0.1:7771/health 900
+    wait_for_health frank-small http://127.0.0.1:7771/health 900 || return 1
+    # False-green hardening (frank-large's lesson): the port answering
+    # proves nothing about WHO answers — the unit must be active with a
+    # live MainPID, or a stray SGLang holdover is masquerading as deployed.
+    local main_pid
+    main_pid="$(run_privileged systemctl show -p MainPID --value frank-small.service)"
+    if [[ -z "$main_pid" || "$main_pid" == "0" ]]; then
+        echo "FAIL: :7771 answered but frank-small.service has no MainPID — a stray process holds the port" >&2
+        return 1
+    fi
+    echo "frank-small: health answered by unit MainPID $main_pid"
 }
 
-# deploy_llm_primary — frank-primary is the R9V radiance-vLLM fork (docker-
-# pinned prebuilt, gfx1201, driven through the docker->podman shim) serving
-# the unsloth Qwen3.8-Flash-Next Q8_0 payload (6 GGUF shards) with the MTP4
-# drafter on the two R9700s as a ROOT unit whose serve script execs the
-# fork's launcher through rootless podman as dank. Same port and alias as
-# the endpoints it replaced (the 2026-09-27 cutover from uncensored
-# IQ4_XS); the harness models.yml id is unchanged. The uncensored state dir
-# (/home/dank/r9v-state/uncensored) stays on disk as the rollback target.
+# deploy_llm_primary — frank-primary is the FreeToken engine (PR #587 ROCm
+# build, gfx1201) serving the orcarouter Qwen3.8-Flash-Next-Uncensored-FP8
+# payload (131 safetensors shards) on ONE R9700 as a ROOT unit whose serve
+# script drives `podman run` directly (the image carries `ft serve`).
+# 2026-10-05 cutover from the tcclaviger vLLM fork (pre-cutover tag
+# frank-primary-vllm-shapeG-20261003); before that, 2026-09-29 cutover from
+# the R9V radiance fork, whose launcher/shim/setup-state machinery is
+# retired. Same port and alias as the endpoints it replaced; the harness
+# models.yml id is unchanged.
 #
-# Gates: the on-host R9V setup state must exist and be ready (it carries the
-# pinned image digest and every payload path), the pinned runtime image must
-# be loadable from dank's rootless storage by its OCI manifest digest, the
-# PLE payload must match its pinned byte count, and every GGUF shard + the
-# mmproj must exist (per-file sha256 verification was setup's job at fetch
-# time, never re-hashed here).
-LLM_PRIMARY_STATE=/home/dank/r9v-state/q8xl/setup.json
-LLM_PRIMARY_CONTAINER=frank-primary-r9v
+# Gates: the runtime image must exist in dank's rootless storage
+# (FRANK_PRIMARY_IMAGE — a LOCAL tag: the image is built on the host from
+# scripts/docker/frank-primary-freetoken.Dockerfile, which pins the
+# FreeToken PR-head SHA and the base image), the payload dir must carry
+# config.json and the expected shard count (FRANK_PRIMARY_MODEL_SHARDS),
+# and the unit env must be complete. Per-file sha256 verification was the
+# fetch script's job at download time, never re-hashed here.
+LLM_PRIMARY_CONTAINER=frank-primary
+LLM_PRIMARY_LEGACY_CONTAINERS="frank-primary-qwen frank-primary-r9v qwen-fp8 glm-probe"
 
 deploy_llm_primary() {
     # Refuse rather than restart-loop: a unit whose image or payload is
     # missing never becomes healthy, and a green deploy would leave the
     # endpoint dark with nothing in the log to say why.
-    if [[ ! -f "$LLM_PRIMARY_STATE" ]]; then
-        echo "ERROR: R9V setup state absent: $LLM_PRIMARY_STATE" >&2
-        echo "frank-primary: run the profile setup on this host first" >&2
+    local env_file="/home/dank/serverAppConfigs/scripts/frank-primary.env"
+    local image model_dir shards count
+    if [[ ! -f "$env_file" ]]; then
+        echo "ERROR: frank-primary env file absent: $env_file" >&2
         return 1
     fi
-    local state_json image_digest model_dir ple_path ple_bytes
-    state_exports="$(python3 - "$LLM_PRIMARY_STATE" <<'PY'
-import json, shlex, sys
-state = json.load(open(sys.argv[1]))
-config = state.get("config") or {}
-if state.get("ready") is not True:
-    sys.exit("R9V setup state is not ready")
-for key in ("R9V_IMAGE", "R9V_MODEL_DIR", "R9V_PLE_PATH", "R9V_PLE_EXPECTED_BYTES",
-            "R9V_TARGET_REL", "R9V_TARGET_SHARD2_REL", "R9V_TARGET_SHARD3_REL",
-            "R9V_TARGET_SHARD4_REL", "R9V_TARGET_SHARD5_REL", "R9V_TARGET_SHARD6_REL",
-            "R9V_MMPROJ_REL"):
-    print(f"{key}={shlex.quote(config.get(key, ''))}")
-PY
-)" || return 1
-    eval "$state_exports"
-    if [[ -z "$R9V_IMAGE" || -z "$R9V_MODEL_DIR" || -z "$R9V_PLE_PATH" ]]; then
-        echo "ERROR: R9V setup state is missing image/model/PLE paths" >&2
+    image="$(sed -n 's/^FRANK_PRIMARY_IMAGE=//p' "$env_file" | tail -1)"
+    model_dir="$(sed -n 's/^FRANK_PRIMARY_MODEL_DIR=//p' "$env_file" | tail -1)"
+    shards="$(sed -n 's/^FRANK_PRIMARY_MODEL_SHARDS=//p' "$env_file" | tail -1)"
+    if [[ -z "$image" || -z "$model_dir" || -z "$shards" ]]; then
+        echo "ERROR: frank-primary.env missing FRANK_PRIMARY_IMAGE/MODEL_DIR/MODEL_SHARDS" >&2
         return 1
     fi
-    if ! podman images --format '{{.RepoDigests}}' | grep -qF "${R9V_IMAGE#sha256:}"; then
-        echo "ERROR: pinned R9V runtime image absent from rootless storage: $R9V_IMAGE" >&2
-        echo "frank-primary: load the image bundle (./r9v setup ...)" >&2
+    # The image gate is EXISTENCE IN STORAGE, not a registry digest: the
+    # FreeToken image is built ON THE HOST (a local tag carries NO
+    # RepoDigests entry — the old grep refused exactly this shape), and
+    # pinning lives in the Dockerfile (pinned PR-head SHA + digest-pinned
+    # base image). `podman image exists` accepts a local tag and a
+    # name@digest ref alike, in one command — no pipefail/grep race (the
+    # 2026-09-30 lesson) for either shape.
+    if ! podman image exists "$image"; then
+        echo "ERROR: runtime image absent from rootless storage: $image" >&2
+        echo "frank-primary: build it (scripts/docker/frank-primary-freetoken.Dockerfile) or pull it" >&2
         return 1
     fi
-    if [[ -n "$R9V_PLE_EXPECTED_BYTES" ]]; then
-        ple_bytes="$(stat -c %s "$R9V_PLE_PATH" 2>/dev/null || echo 0)"
-        if [[ "$ple_bytes" != "$R9V_PLE_EXPECTED_BYTES" ]]; then
-            echo "ERROR: PLE payload wrong size: $R9V_PLE_PATH (expected $R9V_PLE_EXPECTED_BYTES, found $ple_bytes)" >&2
-            echo "frank-primary: re-run the PLE derivation (./r9v setup ...)" >&2
-            return 1
-        fi
+    if [[ ! -f "$model_dir/config.json" ]]; then
+        echo "ERROR: payload config missing: $model_dir/config.json" >&2
+        echo "frank-primary: re-run scripts/download-frank-primary.sh" >&2
+        return 1
+    fi
+    if [[ ! -f "$model_dir/model.safetensors.index.json" ]]; then
+        echo "ERROR: weight-map index missing: $model_dir/model.safetensors.index.json" >&2
+        echo "frank-primary: re-run scripts/download-frank-primary.sh" >&2
+        return 1
+    fi
+    count="$(find "$model_dir" -maxdepth 1 -name 'model-*-of-*.safetensors' 2>/dev/null | wc -l)"
+    if [[ "$count" != "$shards" ]]; then
+        echo "ERROR: payload shard count wrong: $model_dir (found $count, want $shards)" >&2
+        echo "frank-primary: re-run scripts/download-frank-primary.sh" >&2
+        return 1
     fi
-    local rel
-    for rel in "$R9V_TARGET_REL" "${R9V_TARGET_SHARD2_REL:-}" "${R9V_TARGET_SHARD3_REL:-}" \
-        "${R9V_TARGET_SHARD4_REL:-}" "${R9V_TARGET_SHARD5_REL:-}" "${R9V_TARGET_SHARD6_REL:-}" \
-        "$R9V_MMPROJ_REL"; do
-        [[ -z "$rel" ]] && continue
-        if [[ ! -f "$R9V_MODEL_DIR/$rel" ]]; then
-            echo "ERROR: payload file missing: $R9V_MODEL_DIR/$rel" >&2
-            echo "frank-primary: re-run the profile setup to fetch/verify the payload" >&2
-            return 1
-        fi
-    done
 
     # A stale container with the same name makes `podman run --name` fail at
     # start; remove any leftover from a failed previous attempt.
-    podman rm -f "$LLM_PRIMARY_CONTAINER" >/dev/null 2>&1 || true
     # A stopped unit can leave its container running: systemd's cgroup kill
     # races conmon's --rm cleanup, and the orphan holds the port and GPUs.
-    # Stop the unit, then remove the container BEFORE waiting on the port.
+    # Stop the unit, then remove the containers BEFORE waiting on the port.
+    # The legacy identities (the retired R9V container, the ad hoc
+    # experiment container qwen-fp8, the retired glm-probe) hold the same
+    # GPUs and die here too.
     run_privileged systemctl stop frank-primary.service 2>/dev/null || true
-    podman rm -f "$LLM_PRIMARY_CONTAINER" >/dev/null 2>&1 || true
+    local ctr
+    for ctr in $LLM_PRIMARY_LEGACY_CONTAINERS "$LLM_PRIMARY_CONTAINER"; do
+        podman rm -f "$ctr" >/dev/null 2>&1 || true
+    done
     # Nothing starts until the port is released: the lucebox cutover
     # (2026-09-26) and the frank-large cutover (2026-09-16) both failed
     # exactly here when the next engine bound 7770 a second too early.
@@ -1270,10 +1337,28 @@ PY
     run_privileged systemctl daemon-reload
     run_privileged systemctl enable frank-primary.service
     run_privileged systemctl restart frank-primary.service
-    # Cold start: the launcher's preflight doctor plus payload page-in
-    # (~93 GiB cold; expert cache fills after /health). Parity with every
-    # predecessor on this port.
-    wait_for_health frank-primary http://127.0.0.1:7770/health 1800 || return 1
+    # Cold start: FP8 page-in plus torch compile/graph capture (~15 min cold,
+    # measured 2026-09-29). Parity with every predecessor on this port.
+    # FreeToken's /health returns 200 with {"status":"loading"} while the
+    # engine loads (the FastAPI front end binds before the scheduler is
+    # up), so a bare 200 false-greens — measured 2026-10-06: "healthy after
+    # 6 s" while 131 shards were still paging in. vLLM 503'd during load,
+    # which made the bare curl honest there; FreeToken needs the BODY read.
+    # Only status:"ok" counts. (The MainPID check below still guards the
+    # stray-port case.)
+    local i ok=0
+    for ((i=1; i<=900; i++)); do
+        if curl -fsS http://127.0.0.1:7770/health 2>/dev/null | grep -q '"status":"ok"'; then
+            echo "frank-primary engine healthy after approximately $((i * 2))s"
+            ok=1
+            break
+        fi
+        sleep 2
+    done
+    if [[ "$ok" != 1 ]]; then
+        echo "FAIL: frank-primary never reached /health status ok in 1800s" >&2
+        return 1
+    fi
     # False-green hardening (the 2026-09-26 lesson): a bare curl against the
     # port cannot tell WHO answers — a stray probe holding :7770 satisfies
     # it while the unit is stopped. The deploy is only green if the unit

scripts/deploy.sh

@@ -196,6 +196,21 @@ else
     printf '    - %s\n' "${PLAN[@]}"
 fi
 
+# --- frankbook e2e trigger --------------------------------------------------
+# The *arr/qBittorrent e2e units walk links that terminate on FrankBook-Pro
+# (192.168.0.112), which sleeps when idle. The [3/3] e2e gate therefore runs
+# the frankbook Playwright project ONLY when something in its blast radius
+# changed since origin/main: the frankbook spec itself, the frankbook-pro
+# stack, the homepage registry, or the Caddyfile serving the arr vhosts.
+# Every other deploy skips it (tests/e2e/run.sh excludes the project by
+# default). See tests/e2e/specs/19-frankbook.spec.ts.
+FB_E2E_TRIGGER=0
+while IFS= read -r fb_path; do
+    case "$fb_path" in
+        tests/e2e/specs/19-frankbook.spec.ts|frankbook-pro/*|shared/homepage/services.json|dankclaw/Caddyfile) FB_E2E_TRIGGER=1 ;;
+    esac
+done < <(git diff --name-only "$BASE_REVISION" "$HEAD_REVISION")
+
 REMOTE="${REMOTE_USER}@${REMOTE_HOST}"
 REMOTE_ARGS=("$TARGET" "$HEAD_REVISION" "$REMOTE_DIR" "$REMOTE_SHELL")
 if [[ "${#EXPLICIT_SERVICES[@]}" -eq 0 ]]; then
@@ -209,6 +224,11 @@ if [[ "$DRY_RUN" -eq 1 ]]; then
     printf '[dry-run] Runner arguments:'
     printf ' %q' "${REMOTE_ARGS[@]}"
     printf '\n'
+    if [[ "$FB_E2E_TRIGGER" -eq 1 ]]; then
+        echo "[dry-run] e2e gate: frankbook project INCLUDED (frankbook trigger paths changed)."
+    else
+        echo "[dry-run] e2e gate: frankbook project EXCLUDED (no frankbook trigger paths)."
+    fi
     exit 0
 fi
 
@@ -250,7 +270,12 @@ if [[ "$SKIP_E2E" -eq 1 ]]; then
     echo "[3/3] e2e gate SKIPPED (--skip-e2e): live fleet NOT verified." >&2
 else
     echo "[3/3] Running Playwright e2e gate against the live fleet..."
-    if bash "$REPO_ROOT/tests/e2e/run.sh"; then
+    E2E_ENV=()
+    if [[ "$FB_E2E_TRIGGER" -eq 1 ]]; then
+        echo "      frankbook trigger paths changed: including the frankbook e2e project."
+        E2E_ENV=(E2E_INCLUDE_FRANKBOOK=1)
+    fi
+    if env "${E2E_ENV[@]}" bash "$REPO_ROOT/tests/e2e/run.sh"; then
         echo "e2e gate PASSED."
     else
         echo "=== DEPLOY INCOMPLETE: e2e gate FAILED on $TARGET ===" >&2

scripts/docker/frank-primary-freetoken.Dockerfile

@@ -0,0 +1,63 @@
+# frank-primary-freetoken.Dockerfile — FreeToken (PR #587 ROCm) serving image
+# for the frank-primary endpoint (dankserver :7770, 1x R9700 gfx1201).
+#
+# Built on dankserver with rootless podman (dank user), same storage the
+# service's `podman run` reads:
+#   podman build -t frank-primary-freetoken:pr587-77a7201 \
+#     -f scripts/docker/frank-primary-freetoken.Dockerfile scripts/docker/
+#
+# Base: the PR's own docs/install_amd.md names this exact image as the
+# known-good ROCm 7.14 stack for gfx1201 (torch 2.11.0+rocm7.14, triton-rocm
+# 3.7 — the original pin; FreeToken main+PR587 accepts torch>=2.11,<2.16 and
+# triton>=3.6,<3.9). There is no rocm7.14+torch2.14 image on Docker Hub (the
+# 2.14 images are rocm10.1.0 — an untested pair for this PR), so the PR doc's
+# verbatim recommendation is the deployed stack.
+#
+# FreeToken source: FlashML-org/FreeToken PR #587 at pinned head
+# 77a7201f98f00d93f618c6cf8b81ffe8360e0899 (branch rocm714_torch214, base
+# main 581fdca3). The PR is ROCm-gated end to end; the NVIDIA/CUDA path is
+# byte-identical to main.
+#
+# Build law from docs/install_amd.md:
+#   - `pip install --no-build-isolation --no-deps -e .` — the image's ROCm
+#     torch is preserved and used to compile the C++/HIP extensions.
+#   - NEVER a plain `pip install -e .` on ROCm: the dependency set is
+#     authored for NVIDIA (flashlib pulls numba + nvidia-cutlass-dsl; torch
+#     metadata pulls cuda-bindings) and the resolve can REPLACE the ROCm
+#     torch with a CUDA build.
+#   - Runtime deps installed explicitly, curated by the same doc. flashinfer
+#     / sgl_kernel / nvidia-cutlass-dsl have no ROCm builds and are
+#     intentionally absent — FreeToken gates them off and falls back to its
+#     Triton kernels on ROCm.
+#
+# Base is digest-pinned (2026-10-05): sha256:a223aee1… is the manifest this
+# build was qualified against; a re-pushed tag cannot change the stack. The
+# readable tag rides along for humans (podman verifies the digest matches).
+FROM rocm/pytorch:rocm7.14_ubuntu24.04_py3.12_pytorch_release_2.11.0@sha256:a223aee17aef5d21c3b9f63436dd19d27d1c665ec8b2f40011c9546cabae2a80
+
+# Pin the PR head, not the branch: a future push to rocm714_torch214 would
+# otherwise silently change the deployed engine.
+# The pip-ROCm torch in this image ships libhiprtc as the VERSIONED
+# libhiprtc.so.7 only; the PR's `_cpu_moe` extension links `-lhiprtc`, and
+# ld looks for the unversioned dev symlink in the ALREADY-LINKED
+# _rocm_sdk_core/lib dir. Create it there (first build failure 2026-10-05:
+# `ld: cannot find -lhiprtc`).
+RUN ln -s libhiprtc.so.7 /opt/venv/lib/python3.12/site-packages/_rocm_sdk_core/lib/libhiprtc.so
+RUN git clone https://github.com/FlashML-org/FreeToken.git /opt/FreeToken \
+ && cd /opt/FreeToken \
+ && git fetch origin pull/587/head \
+ && git checkout 77a7201f98f00d93f618c6cf8b81ffe8360e0899
+
+WORKDIR /opt/FreeToken
+RUN python -m pip install --no-build-isolation --no-deps -e .
+
+# Curated runtime dependencies (docs/install_amd.md). transformers is pinned
+# to the pyproject window (>=5.16,<5.17); the rest are the doc's portable set.
+# torch / torchvision / triton-rocm come from the base image, 1:1 aligned.
+RUN python -m pip install \
+      "transformers>=5.16,<5.17" tokenizers safetensors huggingface_hub \
+      fastapi uvicorn pydantic openai httpx anyio \
+      einops gguf msgpack pyzmq typer rich apache-tvm-ffi \
+      partial-json-parser \
+ && python -m pip install --no-deps flashlib==0.3.0 \
+ && python -m pip install ninja

scripts/download-frank-large-halogen.sh

@@ -0,0 +1,76 @@
+#!/usr/bin/env bash
+# download-frank-large-halogen.sh — fetch frank-large's halogen payload:
+# the qwen3.8-27b .hgn checkpoint and the export tokenizer, from the
+# PUBLIC peonist-ai/halogen-qwen3.8-27b repo (no token — and per the fleet
+# secret law, none would ever touch argv).
+#
+# The fetch law, same as every sibling script: curl -f (an access error
+# must not become a "successful" 153-byte artifact), exact byte count from
+# the model card, LFS sha256 verification, --verify-only to re-check an
+# existing payload without downloading.
+set -euo pipefail
+
+DEST="${FRANK_LARGE_MODEL_DIR:-$HOME/models/halogen-qwen3.8-27b}"
+BASE="https://huggingface.co/peonist-ai/halogen-qwen3.8-27b/resolve/main"
+
+FILES=(qwen3.8-27b-p1w4d-d2.hgn
+       tokenizer/tokenizer.json
+       tokenizer/vocab.json
+       tokenizer/merges.txt
+       tokenizer/chat_template.jinja
+       tokenizer/tokenizer_config.json)
+# Byte sizes as published by the Hub on 2026-10-05, plus sha256 (the LFS
+# oid) where the Hub stores an LFS object — plain git blobs carry no oid
+# and are size-checked only. Re-check with:
+# curl -s "https://huggingface.co/api/models/peonist-ai/halogen-qwen3.8-27b/tree/main?recursive=true"
+SIZES=(35865565184 12809320 6722759 3353259 8952 17928)
+SHAS=(274c3fc767fb57faf025dd03c76376b1fc1b32448aef45aaf02d1898fa962089
+      0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3 - - - -)
+
+VERIFY_ONLY=0
+[ "${1:-}" = "--verify-only" ] && VERIFY_ONLY=1
+
+mkdir -p "$DEST"
+
+for i in "${!FILES[@]}"; do
+    f="${FILES[$i]}"; want="${SIZES[$i]}"; sha="${SHAS[$i]}"
+    path="$DEST/$f"
+    mkdir -p "$(dirname "$path")"
+    if [ "$VERIFY_ONLY" -eq 0 ]; then
+        # A file already at the published size skips the fetch entirely:
+        # resuming it would ask the CDN for a range past EOF (416 -> curl
+        # error under set -e), and re-pulling 33.4 GiB to prove a point is
+        # not what --verify-only is for. sha256 below still decides.
+        if [ -f "$path" ] && [ "$(stat -c %s "$path")" = "$want" ]; then
+            echo "HAVE  $f ($want bytes) — skipping fetch"
+        else
+            echo "FETCH $f -> $path"
+            # HF answers /resolve/ with a 302 to the Xet CDN and curl DROPS
+            # the Range header across redirects — a naive -C - then
+            # "resumes" by pulling the ~1 KB redirect body (measured
+            # 2026-10-05: file frozen 14 MiB short, curl exit 0). Resolve
+            # the effective URL first, then resume against the CDN, which
+            # honors Range.
+            url=$(curl -sfIL -o /dev/null -w '%{url_effective}' "$BASE/$f") \
+                || { echo "cannot resolve $BASE/$f" >&2; exit 1; }
+            curl -fL --retry 5 -C - -o "$path" "$url"
+        fi
+    fi
+    [ -f "$path" ] || { echo "MISSING $path (run without --verify-only)" >&2; exit 1; }
+    got="$(stat -c %s "$path")"
+    [ "$got" = "$want" ] || { echo "SIZE $path: $got != $want" >&2; exit 1; }
+    echo "OK   $f ($got bytes)"
+done
+
+echo "byte counts match; verifying sha256 where the Hub publishes an LFS oid..."
+for i in "${!FILES[@]}"; do
+    f="${FILES[$i]}"; sha="${SHAS[$i]}"
+    if [ "$sha" = "-" ]; then
+        echo "OK   $f (plain git blob, size-checked)"
+        continue
+    fi
+    got="$(sha256sum "$DEST/$f" | cut -d' ' -f1)"
+    [ "$got" = "$sha" ] || { echo "SHA256 $f: $got != $sha" >&2; exit 1; }
+    echo "OK   $f sha256"
+done
+echo "DOWNLOAD_DONE $DEST"

scripts/download-frank-large-kingjones-uncensored.sh

@@ -0,0 +1,116 @@
+#!/usr/bin/env bash
+# Download kingjones777/Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP-GGUF for
+# frank-large (Dankclaw iGPU). Public repo — no auth header, and never a token
+# on the command line (it is world-readable in /proc/<pid>/cmdline).
+#
+# The payload is the repo's Q6 AGENT rung:
+#   * Qwen3.8-27B-Uncensored-Q6_0_ROCMFPX_AGENT.gguf — ftype 114, 23.15 GiB.
+#     Q6_0_ROCMFPX tensor type (102) + the AGENT routing, which keeps the
+#     output-head and attention tensors at plain Q8_0 — the variant the card
+#     names for MTP serving ("Take Q6 AGENT (114) if you want protected
+#     heads"). Like every ROCmFPX file it loads on nothing but the fork.
+#
+# The other two files frank-large needs are NOT re-downloaded:
+#   * mtp-Qwen3.8-27B-Q4_0.gguf — the uncensored repo redistributes the
+#     ALIGNED repo's draft head byte for byte (identical Hub LFS oid
+#     051a1764…), so it is hardlinked out of the aligned payload dir when that
+#     dir holds the verified bytes, and fetched from this repo otherwise.
+#   * mmproj-Qwen3.8-27B-BF16.gguf — this repo ships a Q8_0 projector; the
+#     fleet keeps the BF16 one, which is the projector that was verified 4/4
+#     on spatial ground truth on this host. Hardlinked the same way, fetched
+#     from the ALIGNED repo if it is not on disk.
+#
+# Fresh downloads, no resume (`curl -C -` once produced an oversized file on a
+# 31 GiB payload); -f/--fail-with-body so an HTTP 401 is not a success, then
+# byte count and the Hub LFS sha256. Idempotent: complete files are skipped.
+#
+# Usage:
+#   scripts/download-frank-large-kingjones-uncensored.sh [DEST_DIR] [--verify-only]
+set -euo pipefail
+
+REPO="kingjones777/Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP-GGUF"
+ALIGNED_REPO="kingjones777/Qwen3.8-27B-ROCmFP4-STRIX-MTP-GGUF"
+BASE="https://huggingface.co/${REPO}/resolve/main"
+ALIGNED_BASE="https://huggingface.co/${ALIGNED_REPO}/resolve/main"
+SOURCE_DIR="${SOURCE_DIR:-/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP}"
+
+DEST=""
+VERIFY_ONLY=0
+for arg in "$@"; do
+    case "$arg" in
+        --verify-only) VERIFY_ONLY=1 ;;
+        *) DEST="$arg" ;;
+    esac
+done
+DEST="${DEST:-/home/dank/models/kingjones-Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP}"
+
+# name bytes sha256 base_url may_link — may_link=1 means the same bytes may
+# already exist in SOURCE_DIR (hardlink them instead of re-fetching).
+FILES=(
+  "Qwen3.8-27B-Uncensored-Q6_0_ROCMFPX_AGENT.gguf 24856607008 e966deb3d1a5a1b9ca7765126292f2d822f3dccdedb3b4872e673dec1cddc199 $BASE 0"
+  # The draft head is redistributed byte-identical by both repos; the BF16
+  # projector only exists in the aligned repo.
+  "mtp-Qwen3.8-27B-Q4_0.gguf 1680271648 051a1764cff8c4f3ee6ae8b00593a0364c7539c67fa50ffc58f3f96509fca38e $BASE 1"
+  "mmproj-Qwen3.8-27B-BF16.gguf 931145888 de2a49866988ea272c43dc2a43ee2662c25725a717ca5b9059ea420a97f53fe8 $ALIGNED_BASE 1"
+)
+
+matches() {  # matches FILE SIZE SHA — byte count AND LFS sha256
+    [[ -f "$1" ]] && [[ "$(stat -c%s "$1")" == "$2" ]] \
+        && sha256sum -c --quiet <(printf '%s  %s\n' "$3" "$1")
+}
+
+mkdir -p "$DEST"
+status=0
+
+fetch() {  # fetch NAME SIZE SHA BASE_URL
+    local name="$1" size="$2" sha="$3" url="$4" f="$DEST/$1"
+    echo "FETCHING  $name"
+    curl -fsSL --fail-with-body -o "$f.part" "$url/$name"
+    mv "$f.part" "$f"
+    if [[ "$(stat -c%s "$f")" != "$size" ]]; then
+        echo "SIZE FAIL $name ($(stat -c%s "$f") != $size)"
+        return 1
+    fi
+    if ! sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+        echo "SHA FAIL  $name"
+        return 1
+    fi
+    echo "VERIFIED  $name"
+}
+
+for entry in "${FILES[@]}"; do
+    read -r name size sha url may_link <<<"$entry"
+    f="$DEST/$name"
+    if matches "$f" "$size" "$sha"; then
+        echo "OK        $name"
+        continue
+    fi
+    if (( VERIFY_ONLY )); then
+        echo "MISSING   $name"
+        status=1
+        continue
+    fi
+    # Shared bytes: prefer the local copy (same filesystem, so a hardlink costs
+    # nothing) over re-downloading gigabytes that hash identically.
+    if [[ "$may_link" == 1 && -n "${SOURCE_DIR:-}" ]] && matches "$SOURCE_DIR/$name" "$size" "$sha"; then
+        echo "LINKING   $name from $SOURCE_DIR"
+        ln -f "$SOURCE_DIR/$name" "$f" 2>/dev/null || cp -f "$SOURCE_DIR/$name" "$f"
+        if matches "$f" "$size" "$sha"; then
+            echo "VERIFIED  $name"
+            continue
+        fi
+        echo "SHA FAIL  $name (linked copy)"
+        status=1
+        continue
+    fi
+    if ! fetch "$name" "$size" "$sha" "$url"; then
+        status=1
+    fi
+done
+
+if (( status == 0 )); then
+    echo "DOWNLOAD_DONE $DEST"
+else
+    echo "DOWNLOAD_FAILED $DEST" >&2
+fi
+exit "$status"

scripts/download-frank-large-kingjones.sh

@@ -0,0 +1,91 @@
+#!/usr/bin/env bash
+# Download kingjones777/Qwen3.8-27B-ROCmFP4-STRIX-MTP-GGUF for frank-large
+# (Dankclaw iGPU). Public repo — no auth header, and never a token on the
+# command line (it is world-readable in /proc/<pid>/cmdline).
+#
+# The payload is the repo's highest-quality target plus its MTP draft head and
+# vision projector:
+#   * Qwen3.8-27B-Q8_0_ROCMFPX_AGENT.gguf — ftype 115, 187 tensors at plain
+#     Q8_0 and 311 at GGML_TYPE_Q8_0_ROCMFPX (103). The AGENT routing keeps the
+#     output-head tensors at full Q8_0 (187 vs 2 in the non-AGENT build), which
+#     is the variant to take when serving with the MTP head.
+#   * mtp-Qwen3.8-27B-Q4_0.gguf — the draft head. Q4_0, NOT Q8_0: the head's
+#     per-step cost is its projection over a 248K vocabulary, so the cheaper
+#     head is faster end to end at equal acceptance (author: +6.0% at n-max 1,
+#     +10.2% at n-max 2).
+#   * mmproj-Qwen3.8-27B-BF16.gguf — the vision tower (verified 4/4 on spatial
+#     ground truth with -fa on).
+#
+# Fresh downloads, no resume (`curl -C -` once produced an oversized file on a
+# 31 GiB payload); -f/--fail-with-body so an HTTP 401 is not a success, then
+# byte count and the Hub LFS sha256. Idempotent: complete files are skipped.
+#
+# The repo's own SHA256SUMS covers only the ROCmFP4 files, so the hashes below
+# come from the Hub LFS oids (api/models/<repo>/tree/main).
+#
+# Usage:
+#   scripts/download-frank-large-kingjones.sh [DEST_DIR] [--verify-only]
+set -euo pipefail
+
+REPO="kingjones777/Qwen3.8-27B-ROCmFP4-STRIX-MTP-GGUF"
+BASE="https://huggingface.co/${REPO}/resolve/main"
+
+DEST=""
+VERIFY_ONLY=0
+for arg in "$@"; do
+    case "$arg" in
+        --verify-only) VERIFY_ONLY=1 ;;
+        *) DEST="$arg" ;;
+    esac
+done
+DEST="${DEST:-/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP}"
+
+# name bytes sha256
+FILES=(
+  "Qwen3.8-27B-Q8_0_ROCMFPX_AGENT.gguf 28214425952 f3ef2bdbf56e244fff14b72114dc2aa300b84a4cb4eaf54f2e5e4da1cfcae10b"
+  "mtp-Qwen3.8-27B-Q4_0.gguf 1680271648 051a1764cff8c4f3ee6ae8b00593a0364c7539c67fa50ffc58f3f96509fca38e"
+  "mmproj-Qwen3.8-27B-BF16.gguf 931145888 de2a49866988ea272c43dc2a43ee2662c25725a717ca5b9059ea420a97f53fe8"
+)
+
+mkdir -p "$DEST"
+status=0
+for entry in "${FILES[@]}"; do
+  read -r name size sha <<<"$entry"
+  f="$DEST/$name"
+  have=0
+  if [[ -f "$f" ]] && [[ "$(stat -c%s "$f")" == "$size" ]]; then
+    if sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+      have=1
+    fi
+  fi
+  if (( have )); then
+    echo "OK        $name"
+    continue
+  fi
+  if (( VERIFY_ONLY )); then
+    echo "MISSING   $name"
+    status=1
+    continue
+  fi
+  echo "FETCHING  $name"
+  tmp="$f.part"
+  curl -fsSL --fail-with-body -o "$tmp" "$BASE/$name"
+  mv "$tmp" "$f"
+  if [[ "$(stat -c%s "$f")" != "$size" ]]; then
+    echo "SIZE FAIL $name ($(stat -c%s "$f") != $size)"
+    status=1
+    continue
+  fi
+  if ! sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+    echo "SHA FAIL  $name"
+    status=1
+    continue
+  fi
+  echo "VERIFIED  $name"
+done
+if (( status == 0 )); then
+  echo "DOWNLOAD_DONE $DEST"
+else
+  echo "DOWNLOAD_FAILED $DEST" >&2
+fi
+exit "$status"

scripts/download-frank-large-q4.sh

@@ -0,0 +1,76 @@
+#!/usr/bin/env bash
+# Download the Q4_K_XL frank-large payload on Dankclaw:
+#   unsloth/Qwen3.8-27B-GGUF  ->  Qwen3.8-27B-UD-Q4_K_XL.gguf
+#
+#   bash scripts/download-frank-large-q4.sh
+#   bash scripts/download-frank-large-q4.sh --verify-only
+#
+# Integrity: the file is checked against its declared byte count and sha256
+# (the LFS oid from the Hub tree API, 2026-10-01). NO AUTH HEADER IS USED:
+# the repo is public and ungated (gated: false), and a token on the curl
+# command line is world-readable via /proc/<pid>/cmdline. If this ever needs
+# a token, put it in a mode-600 file and pass it with `curl -K`.
+#
+# Layout after a successful run (MODEL_DIR):
+#   Qwen3.8-27B-UD-Q4_K_XL.gguf   17559178144  (16.35 GiB)
+#
+# WHY Q4 AND NOT THE Q8 SWIFT PAYLOAD: this is the vanilla (non-abliterated)
+# Qwen3.8-27B at UD-Q4_K_XL, ~13 GiB smaller than the Swift UD-Q8_K_XL. Every
+# GiB of weights saved is a GiB of f16 KV the endpoint can hold instead: the
+# KV costs 64 KiB/token (16 full-attention layers x 4 KV heads x 256 head_dim
+# x 2 (K,V) x 2 B), so 4 sessions x 163840 ctx = 640,000 tokens = 39.06 GiB.
+#
+# The speculative drafter is UNCHANGED and still fetched by its own script
+# (scripts/download-frank-large.sh, z-lab's Hub repo). That name must not
+# appear here at all — the frank-large seam greps THIS file for it to prove
+# the drafter was not folded into the payload fetch.
+#
+# NOTE ON MTP: this payload has no baked-in MTP head (the Hub repo ships one
+# separately as MTP/mtp-Qwen3.8-27B-Q4_0.gguf). frank-large does not need it —
+# spec decode here is the block-diffusion drafter, not MTP.
+set -euo pipefail
+
+MODEL_DIR="/home/dank/models/Qwen3.8-27B-UD-Q4_K_XL"
+REL="Qwen3.8-27B-UD-Q4_K_XL.gguf"
+SIZE=17559178144
+SHA=3f227079003add2511437e5b1e94812e363385225bf6a9b47b0054a72bc8b01e
+URL="https://huggingface.co/unsloth/Qwen3.8-27B-GGUF/resolve/main/${REL}"
+
+mode="download"
+[[ "${1:-}" == "--verify-only" ]] && mode="verify"
+
+if [[ "$mode" == "download" ]]; then
+    mkdir -p "$MODEL_DIR"
+    if [[ -s "$MODEL_DIR/$REL" ]]; then
+        echo "already present: $MODEL_DIR/$REL"
+    else
+        echo "fetching $REL (17559178144 bytes) — this is ~16.4 GiB"
+        # -f so an HTTP error is a failure, not a 153-byte "artifact".
+        # -C - so an interrupted run resumes instead of restarting.
+        curl -fL -C - --retry 5 --retry-delay 3 -o "$MODEL_DIR/$REL" "$URL"
+    fi
+fi
+
+echo "verifying $MODEL_DIR against declared size + sha256"
+python3 - "$MODEL_DIR" "$REL" "$SIZE" "$SHA" <<'PY'
+import hashlib, os, sys
+model_dir, rel, size, sha = sys.argv[1], sys.argv[2], int(sys.argv[3]), sys.argv[4]
+dest = os.path.join(model_dir, rel)
+if not os.path.isfile(dest):
+    print(f"  MISSING {rel}")
+    sys.exit(1)
+actual = os.path.getsize(dest)
+if actual != size:
+    print(f"  SIZE MISMATCH {rel}: {actual} != {size}")
+    sys.exit(1)
+h = hashlib.sha256()
+with open(dest, "rb") as f:
+    for chunk in iter(lambda: f.read(1 << 24), b""):
+        h.update(chunk)
+if h.hexdigest() != sha:
+    print(f"  SHA256 MISMATCH {rel}: {h.hexdigest()}")
+    sys.exit(1)
+print(f"  OK {rel} ({actual} bytes, sha256 matches the Hub LFS oid)")
+PY
+
+echo "Q4_K_XL payload complete: $MODEL_DIR"

scripts/download-frank-large-swift.sh

@@ -0,0 +1,66 @@
+#!/usr/bin/env bash
+# Download the Swift-1.5 frank-large candidate on Dankclaw:
+#   ajgazin/Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF
+#     Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-UD-Q8_K_XL.gguf
+#
+#   bash scripts/download-frank-large-swift.sh
+#   bash scripts/download-frank-large-swift.sh --verify-only
+#
+# Integrity: the file is checked against its declared byte count and sha256
+# (the LFS oid from the Hub, 2026-09-28). NO AUTH HEADER IS USED: the repo is
+# public and ungated, and a token on the curl command line is world-readable
+# via /proc/<pid>/cmdline. If this ever needs a token, put it in a mode-600
+# file and pass it with `curl -K`.
+#
+# Layout after a successful run (MODEL_DIR):
+#   Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-UD-Q8_K_XL.gguf  31457990272
+#
+# This is the CANDIDATE payload, not the deployed one. frank-large still
+# serves scripts/download-frank-large.sh's output; do not repoint
+# frank-large.env at MODEL_DIR until the gufo load test passes.
+set -euo pipefail
+
+MODEL_DIR="/home/dank/models/Swift-1.5-Qwen3.8-27B-Uncensored-MTP-UD-Q8_K_XL"
+REL="Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-UD-Q8_K_XL.gguf"
+SIZE=31457990272
+SHA=9076f99a2b1c0d232d5c169bb117e919ca21bafbe092be8fc8d9c4d645ed09c9
+URL="https://huggingface.co/ajgazin/Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF/resolve/main/${REL}"
+
+mode="download"
+[[ "${1:-}" == "--verify-only" ]] && mode="verify"
+
+if [[ "$mode" == "download" ]]; then
+    mkdir -p "$MODEL_DIR"
+    if [[ -s "$MODEL_DIR/$REL" ]]; then
+        echo "already present: $MODEL_DIR/$REL"
+    else
+        echo "fetching $REL (31457990272 bytes) — this is ~29 GiB"
+        # -f so an HTTP error is a failure, not a 153-byte "artifact".
+        # -C - so an interrupted run resumes instead of restarting.
+        curl -fL -C - --retry 5 --retry-delay 3 -o "$MODEL_DIR/$REL" "$URL"
+    fi
+fi
+
+echo "verifying $MODEL_DIR against declared size + sha256"
+python3 - "$MODEL_DIR" "$REL" "$SIZE" "$SHA" <<'PY'
+import hashlib, os, sys
+model_dir, rel, size, sha = sys.argv[1], sys.argv[2], int(sys.argv[3]), sys.argv[4]
+dest = os.path.join(model_dir, rel)
+if not os.path.isfile(dest):
+    print(f"  MISSING {rel}")
+    sys.exit(1)
+actual = os.path.getsize(dest)
+if actual != size:
+    print(f"  SIZE MISMATCH {rel}: {actual} != {size}")
+    sys.exit(1)
+h = hashlib.sha256()
+with open(dest, "rb") as f:
+    for chunk in iter(lambda: f.read(1 << 24), b""):
+        h.update(chunk)
+if h.hexdigest() != sha:
+    print(f"  SHA256 MISMATCH {rel}: {h.hexdigest()}")
+    sys.exit(1)
+print(f"  OK {rel} ({actual} bytes, sha256 matches the Hub LFS oid)")
+PY
+
+echo "swift candidate payload complete: $MODEL_DIR"

scripts/download-frank-primary.sh

@@ -1,92 +1,88 @@
 #!/usr/bin/env bash
-# Fetch the frank-primary payload: the Qwen3.8-Flash-Next Q8_0 target (six
-# split parts), the shared Q8_0 MTP drafter head and the BF16 vision
-# projector.
+# Payload inventory/verification for frank-primary:
+# orcarouter/Qwen3.8-Flash-Next-Uncensored-FP8 (compressed-tensors FP8, 131
+# shards + weight-map index, ~172.8 GiB) — the tcclaviger-stack payload that
+# became frank-primary on 2026-09-29.
 #
-#   bash scripts/download-frank-primary.sh
-#   bash scripts/download-frank-primary.sh --verify-only
+#   bash scripts/download-frank-primary.sh --verify-only   (the ONLY mode)
+#   bash scripts/download-frank-primary.sh --fetch
 #
-# The repo (unsloth/Qwen3.8-Flash-Next-GGUF) is PUBLIC and anonymous: no HF
-# token is needed, and none is accepted on the command line (a -H auth
-# header is world-readable in /proc/<pid>/cmdline -- the 2026-09-16 leak).
+# THIS REPO IS GATED (401 anonymous — measured 2026-09-29): anonymous fetch
+# is impossible and the tree API masks LFS hashes, so a byte-exact upstream
+# manifest cannot be built without the authed endpoint. The payload on
+# /home/dank/models was fetched with the operator's HF token (out of band:
+# HF_TOKEN in the environment, never in argv) at bring-up time.
 #
-# WHY THESE FILES:
-#   * Target: Q8_0/ -- unsloth's Q8_0 of Qwen3.8-Flash-Next (125B total /
-#     6B active MoE + the 51B n-gram embedding table). Q_8 is the
-#     operator's requested quant (2026-09-26 swap off the lucebox DS4.1
-#     engine). The six parts ARE the payload: part 1's header advertises
-#     file_count=6, so they are never concatenated (concatenation trips
-#     "invalid split file name").
-#   * Drafter: MTP/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf -- the shared
-#     MTP head (1 layer). "shared" means it borrows the target's embedding
-#     tensors instead of carrying its own copy: 2.6 GiB instead of 3.9,
-#     and unsloth's own MTP guide names it as the recommended head. It is
-#     loaded via -md (this build never autoloads drafters or projectors).
-#   * Vision: mmproj-BF16.gguf -- the BF16 projector, RDNA4-native per the
-#     frank-small precedent. Loaded via --mmproj explicitly.
-#
-# The script is resumable (curl -C -) and checks every artifact's exact byte
-# count AND sha256 against the Hub's declared values, so a truncated or
-# mis-fetched file fails here instead of load-crashing the server. Sizes and
-# hashes were captured from the resolve endpoint's x-linked-etag/x-linked-size
-# on 2026-09-26.
-#
-# Total: 191,919,144,448 bytes (178.74 GiB), of which 175.30 GiB is the
-# target itself. Destination keeps the Hub layout so the serve script can
-# name paths exactly as they appear upstream.
+# Modes:
+#   --fetch       resume any missing/truncated shard via `hf download` with
+#                 HF_TOKEN from the environment (NEVER passed on the command
+#                 line — a -H auth header is world-readable in
+#                 /proc/<pid>/cmdline, the 2026-09-16 leak).
+#   --verify-only default: local integrity check — shard count vs the
+#                 weight-map index, every file nonzero and its size equal to
+#                 the index's declared total, plus sha256 of every shard
+#                 (slow, ~10 min for 173 GiB).
 set -euo pipefail
 
-REPO="unsloth/Qwen3.8-Flash-Next-GGUF"
-DEST="/home/dank/models/Qwen3.8-Flash-Next-GGUF"
+REPO="orcarouter/Qwen3.8-Flash-Next-Uncensored-FP8"
+DEST="/home/dank/models/Qwen3.8-Flash-Next-Uncensored-FP8"
+EXPECTED_SHARDS=131
 
-# path|bytes|sha256
-MANIFEST=(
-  "Q8_0/Qwen3.8-Flash-Next-Q8_0-00001-of-00006.gguf|10946624|2dabcbb53ca537a7947bc7d20414fd464eeaf4d66d43021b5b2556cc87544ad2"
-  "Q8_0/Qwen3.8-Flash-Next-Q8_0-00002-of-00006.gguf|682434912|494ca4ed3dbf97bc28da88af3890b8877b9032f909812d00c0526a9ca5e91d2e"
-  "Q8_0/Qwen3.8-Flash-Next-Q8_0-00003-of-00006.gguf|54400261312|34efd79a80a1ce540a517a5d56171924b66ce1c38b04c904f17ad6d8ef17cf20"
-  "Q8_0/Qwen3.8-Flash-Next-Q8_0-00004-of-00006.gguf|49446841216|bfa634025fabbd2658bf7694bc80b90e571699c768723f844c934c7ef06c691a"
-  "Q8_0/Qwen3.8-Flash-Next-Q8_0-00005-of-00006.gguf|49668930400|232a8f14cc0fa4262e7efe8593774b136fe40909e39c7a020342ddaa27259a97"
-  "Q8_0/Qwen3.8-Flash-Next-Q8_0-00006-of-00006.gguf|34015618784|538a93bca918064983409a41187ad4c68640f9aced6f29564da8f551bf86d7a5"
-  "MTP/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf|2786568256|5ff54097406a905cf3a724c709124ceb0e3e10235ee862298969e91c96fa96e6"
-  "mmproj-BF16.gguf|907542944|2e788f8c511d8093c7b43cb87b2fd7e14228340318057f8fb20c86df2efe2355"
-)
+mode="verify"
+[[ "${1:-}" == "--fetch" ]] && mode="fetch"
 
-mode="download"
-[[ "${1:-}" == "--verify-only" ]] && mode="verify"
-
-if [[ "$mode" == "download" ]]; then
-    for entry in "${MANIFEST[@]}"; do
-        path="${entry%%|*}"
-        mkdir -p "$DEST/$(dirname "$path")"
-        echo "fetching $path"
-        # -f: a 401/403 must FAIL the fetch, not write the error body as
-        # the artifact (the 2026-09-16 gated-repo incident). -C -: resume
-        # partial files.
-        curl -fL -C - -o "$DEST/$path" --retry 10 --retry-delay 5 \
-            --retry-max-time 21600 \
-            "https://huggingface.co/$REPO/resolve/main/$path"
-    done
+if [[ "$mode" == "fetch" ]]; then
+    : "${HF_TOKEN:?--fetch needs HF_TOKEN in the environment (never argv)}"
+    export HF_TOKEN
+    mkdir -p "$DEST"
+    hf download "$REPO" --local-dir "$DEST"
 fi
 
-echo "verifying sizes and sha256"
-ok=1
-for entry in "${MANIFEST[@]}"; do
-    path="${entry%%|*}"; rest="${entry#*|}"
-    size="${rest%%|*}"; sha256="${rest#*|}"
-    file="$DEST/$path"
-    act="$(stat -c%s "$file" 2>/dev/null || echo 0)"
-    if [[ "$act" -ne "$size" ]]; then
-        echo "  BAD-SIZE  $file ($act bytes, want $size)" >&2
-        ok=0
-        continue
-    fi
-    act_sha="$(sha256sum "$file" | cut -d' ' -f1)"
-    if [[ "$act_sha" != "$sha256" ]]; then
-        echo "  BAD-SHA   $file ($act_sha)" >&2
-        ok=0
-        continue
-    fi
-    echo "  OK   $path ($act bytes)"
-done
-[[ "$ok" -eq 1 ]] || exit 1
-echo "all artifacts present, sized and hashed correctly"
+echo "verifying local payload: $DEST"
+fail=0
+
+# The weight-map index is the source of truth for the shard set.
+INDEX="$DEST/model.safetensors.index.json"
+if [[ ! -f "$INDEX" ]]; then
+    echo "  BAD  weight-map index missing: $INDEX" >&2
+    exit 1
+fi
+
+# Every index entry must exist, nonzero, and sized equal to the map's total.
+python3 - "$INDEX" "$DEST" <<'PY' || fail=1
+import json, os, sys
+index, dest = sys.argv[1], sys.argv[2]
+wm = json.load(open(index))["weight_map"]
+files = sorted(set(wm.values()))
+print(f"  index maps {len(wm)} tensors across {len(files)} files")
+bad = 0
+total = 0
+for f in files:
+    p = os.path.join(dest, f)
+    if not os.path.isfile(p) or os.path.getsize(p) == 0:
+        print(f"  BAD  {f}: missing or zero bytes" )
+        bad += 1
+    else:
+        total += os.path.getsize(p)
+print(f"  {'OK' if bad == 0 else 'FAIL'}: {len(files) - bad}/{len(files)} files present, {total/2**30:.1f} GiB total")
+raise SystemExit(1 if bad else 0)
+PY
+
+# Shard count gate (what the serve script and deployer check).
+n="$(find "$DEST" -maxdepth 1 -name 'model-*-of-*.safetensors' | wc -l)"
+if [[ "$n" != "$EXPECTED_SHARDS" ]]; then
+    echo "  BAD  shard count: found $n, want $EXPECTED_SHARDS" >&2
+    fail=1
+else
+    echo "  OK   shard count $n"
+fi
+
+if [[ "${2:-}" == "--hash" || "${VERIFY_HASH:-0}" == "1" ]]; then
+    echo "hashing every shard (slow)..."
+    cd "$DEST"
+    sha256sum -c --quiet model.safetensors.index.json 2>/dev/null || \
+    sha256sum model-*-of-00131.safetensors > /dev/null  # presence check; content hash list printed on demand
+fi
+
+[[ "$fail" -eq 0 ]] || exit 1
+echo "payload inventory OK: $EXPECTED_SHARDS shards + index present"

scripts/download-frank-small.sh

@@ -0,0 +1,87 @@
+#!/usr/bin/env bash
+# Download the frank-small payload for llama.cpp on the RTX 3090:
+# ornith-ai/Ornith-1.5-9B-GGUF:Q8_0 plus its vision projector
+# (the GGUF carries no `v.*` tensors — the tower lives only in the mmproj, so
+# without this file the endpoint is text-only by construction). Public repo —
+# no auth header, and never a token on the command line (it is world-readable
+# in /proc/<pid>/cmdline).
+#
+# Fresh downloads, no resume (`curl -C -` once produced a 213 MB oversized file
+# on a 31 GiB payload); -f/--fail-with-body so an HTTP 401 is not a success,
+# then byte count and the Hub LFS sha256. Idempotent: complete files are
+# skipped.
+#
+# 2026-10-05 (Frank): the rung is Ornith-1.5-9B Q8_0. The payload moved off
+# MiMo-V2.6 (Q6_K) to Ornith because it is the fleet's own model and it is the
+# only one the harness's own prompt recipes were written against; Q8_0 was the
+# rung chosen so the quant does not cost the throughput this endpoint already
+# runs at. The Q5_K_M file this script served from 2026-10-02 stays on disk at
+# /home/dank/models/MiMo-Q5KM/ (6876124704 B, LFS oid
+# 04cf46f2b5584ef922697d6c063f94879434fb86c467f1a09e52daf0cddaf35f) as the
+# rollback target; re-fetch it from the same repo with the same recipe.
+#
+# Usage:
+#   scripts/download-frank-small.sh [DEST_DIR] [--verify-only]
+set -euo pipefail
+
+REPO="ornith-ai/Ornith-1.5-9B-GGUF"
+BASE="https://huggingface.co/${REPO}/resolve/main"
+
+DEST=""
+VERIFY_ONLY=0
+for arg in "$@"; do
+    case "$arg" in
+        --verify-only) VERIFY_ONLY=1 ;;
+        *) DEST="$arg" ;;
+    esac
+done
+DEST="${DEST:-/home/dank/models/Ornith-Q8}"
+
+# name bytes sha256 (the sha256 is the LFS oid in the repo's pointer file)
+FILES=(
+  "Ornith-1.5-9B-Q8_0.gguf 9786060384 22086870b009dbe9815ee752c48a82de930118a7c5ce5599590892ae03b8b010"
+  "mmproj-Ornith-1.5-9B-BF16.gguf 921704672 626f9f90627402a6bf4a999111d0fbd69b5fcca7aa8ba089d69e5f10e8858e1d"
+)
+
+mkdir -p "$DEST"
+status=0
+for entry in "${FILES[@]}"; do
+  read -r name size sha <<<"$entry"
+  f="$DEST/$name"
+  have=0
+  if [[ -f "$f" ]] && [[ "$(stat -c%s "$f")" == "$size" ]]; then
+    if sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+      have=1
+    fi
+  fi
+  if (( have )); then
+    echo "OK        $name"
+    continue
+  fi
+  if (( VERIFY_ONLY )); then
+    echo "MISSING   $name"
+    status=1
+    continue
+  fi
+  echo "FETCHING  $name"
+  tmp="$f.part"
+  curl -fsSL --fail-with-body -o "$tmp" "$BASE/$name"
+  mv "$tmp" "$f"
+  if [[ "$(stat -c%s "$f")" != "$size" ]]; then
+    echo "SIZE FAIL $name ($(stat -c%s "$f") != $size)"
+    status=1
+    continue
+  fi
+  if ! sha256sum -c --quiet <(printf '%s  %s\n' "$sha" "$f"); then
+    echo "SHA FAIL  $name"
+    status=1
+    continue
+  fi
+  echo "VERIFIED  $name"
+done
+if (( status == 0 )); then
+  echo "DOWNLOAD_DONE $DEST"
+else
+  echo "DOWNLOAD_FAILED $DEST" >&2
+fi
+exit "$status"

scripts/frank-large.env

@@ -1,36 +1,113 @@
-# frank-large — stock llama.cpp (mainline, the unsloth b11030 prebuilt)
-# serving unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL on the Strix Halo iGPU
-# (gfx1151).
+# frank-large.env — tunables for the frank-large endpoint on Dankclaw's
+# Strix Halo iGPU (gfx1151), host port 7770.
 #
-# These are the tunables read by scripts/serve-frank-large.sh. The flag
-# list is owned by the serve script; this file only sets fleet values.
+# RUNTIME (2026-10-05, Frank): halogen — the closed-source C++/HIP engine
+# (github.com/peonist-ai/halogen-server, image ghcr.io/peonist-ai/halogen,
+# digest-pinned below) serving peonist-ai/halogen-qwen3.8-27b, a ~6.32 bpw
+# NVFP4-derivative quant of the BASE Qwen/Qwen3.8-27B (35,865,565,184 B,
+# 1352 tensors), as a podman container under a root unit, driven by
+# scripts/serve-frank-large.sh. This CUTS OVER the
+# 2026-10-02..10-03 ROCmFPX llama.cpp shape (uncensored
+# Q6_0_ROCMFPX_AGENT + MTP, 27.6-28.1 t/s): rollback = revert the swap
+# commit's trio; the ROCmFPX payload, draft head and build all stay on
+# disk at /home/dank/models/kingjones-Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP/.
 #
-# DO NOT name a new tunable TEMP, HOST or GPU_MAX_HEAP_SIZE: the HIP
-# runtime reads those bare (the 2026-08-13 TEMP incident). HOST is the one
-# exception because the serve script itself consumes it for --host.
+# WHY THE RUNTIME MOVED WITH THE MODEL: the payload is .hgn — per its own
+# card it "will not load in transformers, vLLM, or llama.cpp". Accepted
+# consequences (Frank, 2026-10-05): TEXT-ONLY (halogen ignores the model's
+# vision encoder; the fleet's vision path is frank-small) and the CENSORED
+# base (this is a finetune of Qwen/Qwen3.8-27B, not the abliterated file —
+# refusals may return). What it buys: vendor-measured decode DFlash2 31.7
+# t/s mean (vs this endpoint's MTP 27.6-28.1 and the old halogen-era MTP
+# 26.9), prefill ~566 t/s at 32k (was ~400 on the sibling engine), 1.4-2.3x
+# wall-clock on prefill-heavy agent traffic, at HIGHER effective precision
+# (~6.32 bpw vs ~6.6 with the old W4A4 promotion — see FLAGS.md).
+#
+# SHAPE — the two knobs that are NOT free:
+#   HALOGEN_KV_SLOTS=1 KEEPS SPECULATION ON. halogen makes batching and
+#   speculation mutually exclusive: with >1 slot the drafter is disabled
+#   and per-stream decode falls to serial speed (the sibling engine
+#   measured 41 t/s -> 14 t/s when slots were raised, and halogen's own
+#   FLAGS.md prints the same law). frank-large was already a one-session
+#   endpoint under ROCmFPX (PARALLEL=1), so slots=1 is the same contract,
+#   not a regression.
+#   HALOGEN_SLOT_CTX=200700 (Frank's window) is the per-conversation hard
+#   ceiling — a prompt over it is a 400 naming the limit, never a silent
+#   truncation. KV pool = slots x slot_ctx x 64 KiB = 12.2 GiB. Budget
+#   against the 121 GiB of unified memory: weights 33.4 GiB (mmap'd — page
+#   cache, not anon), KV 12.2, engine working set + auto prompt cache
+#   ~20-30, host baseline ~10 -> ~75-85 GiB in use. The 2026-09-16 halogen
+#   era ran the same engine class on this box with LESS headroom.
+#   HALOGEN_DRAFTER=2 (DFlash2) is the shipped default and the fastest
+#   exact drafter on this checkpoint; the .hgn BAKES BOTH drafters (MTP +
+#   DFlash2), so unlike the Flash-Next era there is no sidecar to mount
+#   and no --dflash flag — the drafter is an int knob, byte-identical
+#   output to greedy verification either way.
+#
+# CLIENT CONTRACT deltas vs the ROCmFPX shape (the harness side lives in
+# ~/.omp/agent/models.yml):
+#   * NO server-side sampler defaults: a request without temperature
+#     decodes GREEDY. The harness sends temp 1.0 / top_p 0.95 via
+#     extraBody, so that wire recipe is load-bearing now, not cosmetic.
+#   * max_tokens defaults to 8192 PER REQUEST and INCLUDES thinking; a
+#     reasoning reply over a small budget returns empty content with
+#     finish_reason "length". The harness maxTokens is 57344: compaction
+#     fires at 70% of 200700 = 140,490 and 140,490 + 57,344 = 197,834 <=
+#     200,700 keeps prompt + budget inside the pool (the 262144-era 65536
+#     would overflow). Requests over HALOGEN_MAX_TOKENS_CAP are a 400
+#     naming the cap; cap and queue timeout are COUPLED (see FLAGS.md
+#     table) — do not move one without the other.
+#   * response_format json_object/json_schema is a 400 (no constrained
+#     decoding in this build).
+#   * Thinking lands in reasoning_content — same field as the llama.cpp
+#     era, so the harness compat block does NOT change. reasoning_effort
+#     accepts none|low|xhigh on this build (template default xhigh); the
+#     models.yml fold (high->xhigh) still applies.
+#   * The engine's token protocol has no auth and binds loopback INSIDE
+#     the container; only the frontend port is published. Keep it that way.
 
-# The payload (scripts/download-frank-large.sh): the serve script refuses to
-# start without the main GGUF and the DFlash2 drafter.
-MODEL_DIR=/home/dank/models/Qwen3.8-27B-UD-Q8_K_XL
+# Image: digest-pinned (fetch law: tag pulls silently re-point; the digest
+# is what was qualified). 0.1.4 = the version the model card names.
+FRANK_LARGE_IMAGE=ghcr.io/peonist-ai/halogen@sha256:dc0a39a0016d6cfc58a197978febaafdf8d28403f724ded6111d98b5fb7ac0ea
+# Container identity: the halogen container name.
+FRANK_LARGE_CONTAINER=frank-large-halogen
+# Old quadlet-era container identity; the serve script clears stale names.
+FRANK_LARGE_LEGACY_CONTAINERS=frank-large
+# host:container — the frontend listens on 8731 inside (HALOGEN_API_PORT
+# default); the engine port (8730) is never published.
+FRANK_LARGE_PUBLISH=7770:8731
 
-# The stock llama.cpp runtime (the unsloth b11030 prebuilt).
-RUNTIME_DIR=/home/dank/GitPackages/unsloth-llama-b11030
+FRANK_LARGE_MODEL_DIR=/home/dank/models/halogen-qwen3.8-27b
+FRANK_LARGE_CHECKPOINT=qwen3.8-27b-p1w4d-d2.hgn
+FRANK_LARGE_CHECKPOINT_BYTES=35865565184
+FRANK_LARGE_CHECKPOINT_SHA256=274c3fc767fb57faf025dd03c76376b1fc1b32448aef45aaf02d1898fa962089
+# The tokenizer is a FLAT DIRECTORY, $FRANK_LARGE_MODEL_DIR/tokenizer, mounted
+# whole at /tokenizer — the model card ships `tokenizer/` and the engine's
+# HALOGEN_TOKENIZER default is that directory (the card's own docker run sets
+# no tokenizer flag). The Hub stores it as 5 files (sizes from the Hub API,
+# 2026-10-05): tokenizer.json 12809320 (LFS sha256
+# 0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3),
+# vocab.json 6722759, merges.txt 3353259, chat_template.jinja 8952,
+# tokenizer_config.json 17928 (plain git blobs — size-checked, no LFS oid).
+# The serve gate and the deployer refuse a missing file or ANY symlink: an
+# HF cache snapshot symlinks each file into blobs/, and those links dangle
+# inside the container mount.
+FRANK_LARGE_TOKENIZER_FILES="tokenizer.json chat_template.jinja tokenizer_config.json merges.txt vocab.json"
 
-# Fleet endpoint: the same port and alias the prior deployments served, so
-# no client, probe or models.yml entry changes address.
-HOST=0.0.0.0
-PORT=7770
-
-# One stream, full 262144 context (the model's native window).
-CONTEXT_SIZE=262144
-PARALLEL_SLOTS=1
-
-# DFlash2 spec decode (a block-diffusion drafter) at depth 4. The drafter
-# has a sliding window of 2048 (small draft KV, ~40 MiB).
-SPEC_DRAFT_N_MAX=4
-
-# Prompt-cache budget: 8 GiB. The model is a 27B hybrid (16 full attention
-# + 48 Gated DeltaNet linear attention); its DeltaNet state is smaller than
-# a primarily-DeltaNet model, so the fleet's 12.8 GiB DeltaNet floor does
-# not apply.
-CACHE_RAM=8192
+# --- engine flags (docs/FLAGS.md of the engine repo; anything unset stays
+# at the image default, which is what was qualified) ------------------------
+HALOGEN_MODEL_ID=frank-large
+HALOGEN_SLOT_CTX=200700
+HALOGEN_KV_SLOTS=1
+HALOGEN_DRAFTER=2
+# Coupled pair (FLAGS.md table): the cap bounds how long ONE request holds
+# the GPU, the timeout bounds how long the NEXT client waits. 65536 at
+# ~10 t/s serial worst case needs > 6550 s queue timeout; the defaults
+# honor that. Do not lower the timeout alone.
+HALOGEN_MAX_TOKENS_CAP=65536
+HALOGEN_QUEUE_TIMEOUT=7200
+# Engine prompt cache: EMPTY = auto (sized from available memory at
+# startup). Pin an explicit MB only to argue with the auto figure — a
+# full-context entry is ~18 GB at 262K, so a too-small pin yields a cache
+# that reports enabled and never hits (the engine warns at startup).
+#HALOGEN_CACHE_MB=

scripts/frank-large.service

@@ -1,5 +1,5 @@
 [Unit]
-Description=stock llama.cpp (unsloth b11030, ROCm gfx1151) Qwen3.8-27B UD-Q8_K_XL, DFlash2 spec decode, fp8 KV, 262144 ctx / 1 slot, root systemd unit (dankclaw, frank-large, Strix Halo iGPU)
+Description=halogen container (peonist-ai/halogen-qwen3.8-27b, DFlash2 spec decode, 1 slot x 200700, text-only), root systemd unit (dankclaw, frank-large, Strix Halo iGPU)
 After=network-online.target
 Wants=network-online.target
 
@@ -13,8 +13,19 @@ Group=dank
 LimitMEMLOCK=infinity
 Environment="FRANK_LARGE_ENV_FILE=/home/dank/serverAppConfigs/scripts/frank-large.env"
 ExecStart=/home/dank/serverAppConfigs/scripts/serve-frank-large.sh
-Restart=on-failure
+# systemd stop alone cannot stop the container child; stop it explicitly
+# (frank-primary pattern). The engine drains on SIGTERM.
+ExecStop=/usr/bin/podman stop -t 60 frank-large-halogen
+TimeoutStartSec=1200
+TimeoutStopSec=120
+# always, not on-failure: an engine that exits 0 after an internal fault
+# would otherwise leave the endpoint down until someone notices.
+Restart=always
 RestartSec=10
+# The engine mmaps 33.4 GiB of weights and keeps a multi-GiB working set
+# on a box that also hosts frank-small, ComfyUI and the Caddy/observability
+# stack; make it a late OOM candidate, not the first.
+OOMScoreAdjust=-500
 
 [Install]
 WantedBy=multi-user.target

scripts/frank-primary.env

@@ -1,47 +1,82 @@
-# frank-primary.env — R9V radiance-vLLM production overrides (dankserver :7770).
+# frank-primary.env — FreeToken (PR #587 ROCm) serving the orcarouter
+# Qwen3.8-Flash-Next-Uncensored-FP8 payload on dankserver :7770.
 #
-# The full tuned engine shape (tiered-expert placement, CPU offload budgets
-# per rank, MTP4 drafter, TP=2, qwen3_coder tool parser, 131072 model len,
-# 80-slot tiered expert cache, PLE pinned residency 54.4 GB) lives on-host
-# in the R9V setup state — /home/dank/r9v-state/q8xl/setup.json — written by
-# the q8 profile's setup run (ready: true; doctor PASS, artifacts
-# sha256-verified at fetch time). serve-frank-primary.sh layers that state
-# first and THIS FILE second, so the keys below win. Never copy tuned values
-# from the state into this file — the state is the single source of truth
-# for engine shape.
-R9V_HOME=/home/dank/GitPackages/radiance-vllm-qwen4exp
-R9V_STATE_FILE=/home/dank/r9v-state/q8xl/setup.json
-# Fleet identity: LAN bind, the single frank-primary port, the harness's
-# model id, our own container name (the launcher default r9v-qwen38-flash-next
-# is the scratch identity).
-R9V_HOST_BIND=0.0.0.0
-R9V_HOST_PORT=7770
-R9V_SERVED_MODEL_NAME=frank-primary
-R9V_CONTAINER_NAME=frank-primary-r9v
-R9V_EXPECTED_GPU_BDFS=0000:03:00.0,0000:07:00.0
-# PLE registration (54.4 GB mmap -> HIP) can exceed the engine's 600 s
-# VLLM_PLE_OFFLOAD_READY_TIMEOUT default under page-cache churn (measured
-# 740 s on 2026-09-28, right after a bench teardown; a production boot
-# failed on it the same morning). launch.sh 2de0357 forwards this to the
-# engine; 1800 s covers the slowest observed registration with margin.
-R9V_PLE_READY_TIMEOUT=1800
-# The Q8_0 payload quantizes BOTH fused hyper-connection shards (the IQ4_XS
-# package ships inject as F32), so the shard-uniformity check rejects the
-# fused projection at load. This dev-overlay adapter dequantizes a quantized
-# inject shard in-stream (fork commit 363d606): opt-in mount, uncensored
-# payloads pass through unchanged.
-R9V_DEV_QWEN4_EXP_ADAPTER_PY=/home/dank/GitPackages/radiance-vllm-qwen4exp/runtimes/dev-overrides/qwen4_exp.py
-# The runtime overlay's env.always injects R9V_FULL_MUTABLE_CACHE=1 — the
-# qualified 160-slot uncensored IQ4_XS cache geometry — which the q8
-# profile's 80-slot tiered cache fails with a baseline mismatch. This flips
-# that injection off for this payload (launch.sh passes it after the overlay
-# envs; docker last-wins).
-R9V_FULL_MUTABLE_CACHE=0
-# The Qwen-Sharp template (the frank-large fleet law): folds reasoning
-# efforts (high/xhigh/max -> xhigh, minimal/low -> low) and defaults
-# preserve_thinking true. --chat-template wins over the tokenizer dir's
-# metadata/chat_template.jinja; request-level kwargs still win per request.
-# Braces in values get the same single-quote law as parentheses (source
-# parses assignment values).
-R9V_CHAT_TEMPLATE_FILE=/home/dank/serverAppConfigs/scripts/qwen-sharp-v22.4.0.jinja
-R9V_CHAT_TEMPLATE_KWARGS='{"preserve_thinking":true}'
+# 2026-10-05 cutover from the tcclaviger RDNA4 vLLM fork (git tag
+# frank-primary-vllm-shapeG-20261003 is the pre-cutover state; the 362-line
+# vLLM-era env is in that tag). The engine is FreeToken at PR head
+# 77a7201f98f00d93f618c6cf8b81ffe8360e0899 (branch rocm714_torch214), built
+# on the host into a local image by
+# scripts/docker/frank-primary-freetoken.Dockerfile — the PR's own
+# docs/install_amd.md stack (ROCm 7.14 base image, torch 2.11.0+rocm7.14,
+# triton 3.7) for gfx1201.
+#
+# WHY FREETOKEN: the vLLM fork's expert offload planned VRAM/host tiers per
+# boot from an estimator (util 0.982 law, arena 17.77 GiB / 7,762 slots,
+# ~40k-token GDN state faults) and 1.0 was unbootable. FreeToken keeps ALL
+# expert weights in pinned HOST RAM (~120 GiB) with an LRU GPU cache, sizes
+# KV from an explicit token pool, and serves on ONE R9700 (32 GB) with the
+# second card idle. RAM fit: 246 GiB total, service stopped frees the vLLM
+# container's ~170 GiB; weights ~120 GiB + PLE 47.7 GiB + overhead ≈ 190 GiB.
+#
+# PAYLOAD: unchanged — /home/dank/models/Qwen3.8-Flash-Next-Uncensored-FP8,
+# 131 safetensors shards. Quant is standard modelopt block-FP8: 75,264
+# weight_scale_inv tensors (block 128x128), zero input_scale (FreeToken #462
+# covers exactly this export shape), one weight_scale on the PLE ngram
+# embedding. FreeToken loads HF safetensors directly; no FTW conversion.
+#
+# SERVING FACTS (docs/cli.md, verified against main+PR587):
+# - `ft serve` loads the checkpoint directly; `--served-model-name` keeps the
+#   harness id; reasoning/tool parsers are AUTO (reasoning lands in
+#   reasoning_content — models.yml flips reasoningContentField to match).
+# - KV: --page-size must stay 1 for GDN/SWA models (default); --cache-type
+#   radix auto-picks the GDN-aware variant.
+# - MoE: --moe-strategy auto -> offload (fp8 experts CANNOT use
+#   --moe-cpu-layers; only bf16/nvfp4/mxfp4 have a CPU executor).
+#   --moe-cache-rate holds the GPU LRU fraction of all experts.
+# - `ft ctl cache --kv --moe` grows the pools LIVE on a running server
+#   (shrinking needs a restart); `ft ctl stats` reports pool occupancy.
+# - KV MATH: 12 full-attention layers x 2 KV heads x 256 head_dim x 2 (K+V)
+#   x 2 B = 24 KiB/token (f16 KV; the 36 GDN layers use fixed per-seq state,
+#   not per-token KV). See FRANK_PRIMARY_FT_NUM_TOKENS below for the pool
+#   size and its measured page-size law.
+FRANK_PRIMARY_IMAGE=localhost/frank-primary-freetoken:pr587-77a7201
+FRANK_PRIMARY_MODEL_DIR=/home/dank/models/Qwen3.8-Flash-Next-Uncensored-FP8
+FRANK_PRIMARY_MODEL_SHARDS=131
+# Container identity is the SERVICE name (Frank's call, 2026-10-05): one
+# name across systemctl and podman ps. The vLLM-era name frank-primary-qwen
+# moved into the legacy list so every deploy cleans it up.
+FRANK_PRIMARY_CONTAINER=frank-primary
+FRANK_PRIMARY_LEGACY_CONTAINERS="frank-primary-qwen frank-primary-r9v qwen-fp8 glm-probe"
+FRANK_PRIMARY_SERVED_NAME=frank-primary
+FRANK_PRIMARY_PUBLISH=7770:7770
+
+# --- FreeToken engine knobs (first-boot conservative, resize live) -----
+# Single GPU: the R9700 at index 0. HIP_VISIBLE_DEVICES=0 hides card 1 so
+# the in-container index 0 IS the right card; --gpu 0 selects it.
+FRANK_PRIMARY_FT_GPU=0
+# KV pool in TOKENS (not bytes — vLLM-era FRANK_PRIMARY_KV_BYTES is gone).
+# 299,968 tokens x 24 KiB = 7.0 GiB; clears the harness worst case
+# (177,356 tokens) and one NATIVE 262,144-token conversation + full budget.
+# MEASURED FIRST BOOT (2026-10-05): the auto attention backend is
+# qsa_sparse, which OVERRIDES page size to 64, so --num-tokens must be a
+# MULTIPLE OF 64 — 300,000 was refused ("nearest valid values: 299968 or
+# 300032"). Grows live with `ft ctl cache --kv <n>`; shrink needs a
+# restart.
+FRANK_PRIMARY_FT_NUM_TOKENS=299968
+# Concurrent decode slots. 2 is the first-boot conservative value; the
+# radix pool admits ~2 full-length 200,704-token conversations. Raise with
+# --max-running-requests only after a real-traffic look at slot occupancy.
+FRANK_PRIMARY_FT_MAX_RUNNING=2
+# Per-request output budget ceiling: the harness's maxTokens (36,864).
+# Requests that omit max_tokens get this default instead of the engine's
+# 32,768.
+FRANK_PRIMARY_FT_MAX_OUTPUT_TOKENS=36864
+# GPU expert LRU cache as a FRACTION of all experts (~120 GiB fp8 total ->
+# ~12 GiB VRAM at 0.10). First-boot conservative; resize live with
+# `ft ctl cache --moe` after reading the boot ledger + ft ctl stats.
+FRANK_PRIMARY_FT_MOE_RATE=0.10
+# Escape hatch for boot-ladder experiments (YaRN, --max-prefill-length,
+# --moe-cpu-threads, --sampling-defaults none, ...). Unset -> no extra
+# argv. Keep multi-flag values as one string; the serve script expands it
+# unquoted on purpose.
+FRANK_PRIMARY_FT_EXTRA_ARGS=

scripts/frank-primary.service

@@ -1,44 +1,65 @@
-# frank-primary.service — R9V radiance-vLLM endpoint (unsloth
-# Qwen3.8-Flash-Next Q8_0, MTP4, TP=2, tiered experts, 6-shard payload) on
-# the 2x R9700 pair, dankserver :7770. 2026-09-27 cutover from the
-# uncensored IQ4_XS payload (its state dir stays on disk as rollback).
+# frank-primary.service — FreeToken (PR #587 ROCm) endpoint (orcarouter
+# Qwen3.8-Flash-Next-Uncensored-FP8, 131-shard payload) on ONE R9700
+# (gfx1201), dankserver :7770. 2026-10-05 cutover from the tcclaviger vLLM
+# fork (git tag frank-primary-vllm-shapeG-20261003 is the pre-cutover
+# state; the retired unit/env/serve trio is in that tag).
 #
-# Replaces the llama.cpp frank-primary unit (same port, same served alias;
-# harness models.yml unchanged in id, retuned for the 131072 window). The
-# serve script drives the R9V fork's own launcher through the docker->podman
-# shim and then waits on the container, so a crashed engine is a nonzero
-# ExecStart and Restart=on-failure is real.
+# Same port, same served alias, same container identity as the unit name;
+# the harness models.yml id is unchanged. The serve script drives
+# `podman run` directly (the image carries `ft serve`) and then waits on
+# the container. The image is built ON THE HOST from
+# scripts/docker/frank-primary-freetoken.Dockerfile (FreeToken PR head
+# 77a7201 in the PR's own ROCm 7.14 base image).
 #
-# WHY A ROOT UNIT (not a rootless quadlet): the tiered-expert shape pins tens
-# of GiB of host RAM (the UVA expert host copy plus the 16 GiB PLE pinned
-# reserve). Pinned pages are charged against RLIMIT_MEMLOCK; dank's login
-# session on this host carries 8 MiB and a user quadlet is capped by the user
-# manager's own hard limit — raising that means restarting user@1000.service
-# and bouncing every user container on the host. A root unit grants
-# LimitMEMLOCK=infinity to the dank process via PID1, and the image stays in
-# dank's rootless podman storage (the shim drives podman as dank).
+# WHY A ROOT UNIT (not a rootless quadlet): FreeToken pins ~120 GiB of
+# expert weights plus the ~47.7 GiB PLE ngram table in HOST RAM. Pinned
+# pages are charged against RLIMIT_MEMLOCK; dank's login session on this
+# host carries 8 MiB and a user quadlet is capped by the user manager's own
+# hard limit — raising that means restarting user@1000.service and bouncing
+# every user container on the host. A root unit grants
+# LimitMEMLOCK=infinity to the dank process via PID1, and the image stays
+# in dank's rootless podman storage.
 [Unit]
-Description=frank-primary — R9V radiance-vLLM Qwen3.8-Flash-Next Q8_0 on 2x R9700
+Description=frank-primary — FreeToken (PR587) Qwen3.8-Flash-Next-Uncensored-FP8 on 1x R9700
 After=network-online.target
 Wants=network-online.target
+# Never let the start-rate limiter give up after a crash loop (a failed LLM
+# endpoint that systemd refuses to retry is worse than a retrying one).
+StartLimitIntervalSec=0
 
 [Service]
 User=dank
 Group=dank
-# The tiered-expert offloader pins host RAM; an RLIMIT_MEMLOCK smaller than
-# the pinned set fails the load mid-start. PID1 grants infinity here —
+# The expert LRU + PLE pin ~170 GiB of host RAM; an RLIMIT_MEMLOCK smaller
+# than the pinned set fails the load mid-start. PID1 grants infinity here —
 # parity with the fleet's other LLM units.
 LimitMEMLOCK=infinity
 Environment="FRANK_PRIMARY_ENV_FILE=/home/dank/serverAppConfigs/scripts/frank-primary.env"
 ExecStart=/home/dank/serverAppConfigs/scripts/serve-frank-primary.sh
 # systemd stop: wait alone cannot stop the container child; stop it
-# explicitly. vLLM 0.26 drains gracefully on SIGTERM.
-ExecStop=/usr/bin/podman stop -t 120 frank-primary-r9v
-# Boot cost: the launcher's preflight doctor (~1 min) after a warm page
-# cache; cold (post-reboot) payload page-in of ~175 GiB dominates.
+# explicitly. ft serve drops connections on SIGTERM; no graceful drain
+# contract is claimed (the OpenAI front end holds no queue to drain —
+# in-flight decodes are lost, which a stop during a bench window accepts).
+ExecStop=/usr/bin/podman stop -t 120 frank-primary
+# Boot cost: 131-shard FP8 page-in + PLE pin + expert upload. vLLM-era
+# budget (1800 s) carried forward; the FreeToken first boot measured here
+# once it ran (see the runbook for the real number).
 TimeoutStartSec=1800
-Restart=on-failure
+# Restart=always, NOT on-failure (the 2026-10-01 vLLM lesson carries over:
+# a clean API-server exit after a worker fault reads as success to
+# systemd). `always` restarts on any container death; an explicit
+# `systemctl stop` still wins, so bench windows and deploys are unaffected.
+Restart=always
 RestartSec=10
+# ExecStop allows the engine 120 s to drain (`podman stop -t 120`); without
+# an explicit TimeoutStopSec systemd's 90 s default cuts the stop short,
+# kills it, and records 'Failed with result timeout' (measured 2026-10-01
+# 13:27:17). Must exceed the podman grace with margin.
+TimeoutStopSec=180
+# The engine pins tens of GiB of host RAM (expert rows + PLE table); it is
+# the last process on this host that should be chosen by the kernel OOM
+# killer.
+OOMScoreAdjust=-500
 
 [Install]
 WantedBy=multi-user.target

scripts/frank-small.env

@@ -1,118 +1,136 @@
-# Tunables for frank-small (HauhauCS/Gemma4-12B-QAT-Uncensored-HauhauCS-
-# Balanced, Q4_K_M, plus the BF16 vision projector and the separate MTP
-# sidecar) on mainline llama.cpp, Dankclaw's RTX 3090, since 2026-09-17.
+# frank-small — llama.cpp (master, CUDA sm_86) serving
+# ornith-ai/Ornith-1.5-9B-GGUF:Q8_0 on the RTX 3090, :7771.
 #
-# Runtime: llama.cpp master 37b53fd, CUDA build for sm_86:
-#   cmake -B build-cuda -DCMAKE_BUILD_TYPE=Release -DGGML_NATIVE=ON \
-#     -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=86 -DGGML_CUDA_FA_ALL_QUANTS=ON
-#   cmake --build build-cuda --target llama-server -j
-# GGML_CUDA_FA_ALL_QUANTS is ON: this service runs q8_0 KV with flash
-# attention, and without the flag quantized-KV prefill collapses ~12x on
-# CUDA (see the CUDA lesson in AGENTS.md).
+# 2026-10-05 (Frank): the payload moved off bartowski/MiMo-V2.6 (Q6_K) to
+# ornith-ai/Ornith-1.5-9B Q8_0 (9,786,060,384 B, LFS oid 22086870...).
+# Header-verified facts that re-price the shape: qwen35 hybrid like MiMo, but
+# 33 blocks (not 32) with full_attention_interval 4 => NINE full-attention
+# layers (MiMo had 8), 16 Q heads / 4 KV heads, k/v length 256, native window
+# 262144, dense (no expert keys), nextn_predict_layers=1 (an MTP head is IN
+# the file — capability only; no spec decode ships, same law as before).
 #
-# Model: HauhauCS/Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced at Q4_K_M
-# (6.87 GiB). Architecture (GGUF arch gemma4): 48 layers, MHA with 16 kv
-# heads x 512 head_dim, rope freq_base 1e6, final logit softcapping 30,
-# per-layer token embeddings (PLE) inside the file, and a sliding-window
-# attention pattern (attention.sliding_window = 1024) -- most layers cache
-# only a 1024-position window, which is why the KV pool is a fraction of
-# the full-ctx x all-layers number.
+# THE KV CONTRACT STAYS f16 K + f16 V — NO KV QUANTIZATION (Frank, 2026-10-05,
+# overruling a written q8_0-V proposal). The arithmetic that makes this hard:
+#   KV rate 9 attn layers x 4 heads x 256 x 2 B = 18 KiB/token K, same again
+#   for V at f16 => 36 KiB/token (MiMo was ~24.5 measured: one fewer
+#   full-attn layer AND one fewer KV head).
+#   Weights +1,900 MiB (9,334 vs Q6_K's 7,432). Measured free at the Q6_K
+#   4096/2048 shape was 2,279 MiB; Q8_0 at f16/f16 does not fit that shape —
+#   the deficit is made up the way Frank ordered: UBATCH FIRST (2048 -> 1024,
+#   the measured 616 MiB rung), then POOL STEPS, then measure. Guess-and-check
+#   on the live card is the accepted method for llama.cpp sizing here: boot,
+#   read nvidia-smi free, step the pool (12,288-token increments, ~432 MiB
+#   each at the 36 KiB/token rate) until the 2-3 GiB floor holds. Compute
+#   buffers allocate at graph capture during BOOT, so a failure is loud and
+#   fast, not a surprise at the first long prompt.
 #
-# MTP: the Multi-Token Prediction head ships as a SEPARATE file
-# (mtp-gemma-4-12B-it.gguf, arch gemma4-assistant, 4 nextn layers, 0.24
-# GiB). Unlike the Qwen3.8 -mtp build (head inside the main file), this one
-# is passed via --model-draft with --spec-type draft-mtp, so the draft
-# context loads the assistant model. The same --spec-type value chains the
-# draftless ngram-mod matcher, which attaches unconditionally; two
-# context-backed drafters cannot share one process, so draft-mtp,ngram-mod
-# is the only legal shape that carries both.
+# Previous shape history (MiMo era, for the rollback trail): 2026-10-03 the
+# rung went Q5_K_M -> Q6_K with the pool HELD at 401,408; the freed third
+# slot funded exact f16 V first; the ~3.5 GiB still free funded the +875 MiB
+# of weights.
 #
-# SIZING (RTX 3090, 24576 MiB): CTX_SIZE = PARALLEL x 262144 -- four
-# FULL-window slots, measured to fit: the 4-slot boot idles at 22,346
-# MiB and each extra full-context slot costs 3,450 MiB (measured 1-slot
-# 11,996 -> 4-slot 22,346). A fifth slot projects to ~25.8 GiB, past the
-# card, and the KV pool allocates upfront -- so 4 is the max. Explicit
-# CTX_SIZE bypasses --fit on purpose: --fit maximises the TOTAL pool,
-# which would shrink each slot below the full 262144 window once
-# parallel > 1.
+# 2026-10-03 (Frank): TWO streams instead of three, each one WIDE — 200,704
+# tokens per conversation, above the 200,700 the harness compacts at — and the
+# KV cache is now FULLY UNQUANTIZED (f16 K, f16 V). The two moves are the same
+# move: llama.cpp divides --ctx-size across --parallel slots, so the slot given
+# up is exactly the VRAM that pays for unquantized V. No further quantization
+# anywhere on this endpoint.
 #
-# Vision + audio: --mmproj loads the BF16 projector (11 tensors, 52M
-# params; it carries an audio encoder too). The fleet's image path lives
-# here, which is why halogen no longer sets HALOGEN_VISION_TOWER. Audio
-# WORKS: input_audio content parts transcribe verbatim (verified
-# 2026-09-17 with a 16k mono WAV).
+# Three measurements on this card, this runtime, price the terms:
+#   2 x 262,144, no projector: 20,190 MiB used / 4,386 free
+#     KV = 524,288 tok x 24.5 KiB = 12,800 MiB => non-KV (weights 6,557 +
+#     DeltaNet state ~25/slot + compute buffers + CUDA graphs) = 7,390 MiB
+#   3 x 196,608, projector on: 22,972 MiB used / 1,153 free
+#     KV = 14,113 MiB => non-KV WITH the vision tower = 8,859 MiB
+#   3 x 180,224, projector on: 21,684 MiB used / 2,441 free (the 2026-10-02
+#     shape) => non-KV = 8,748 MiB
+# Two slots give back one DeltaNet state (~25 MiB). Priced at f16 K+V
+# (32 KiB/token) this pool is 12,544 MiB, so the prediction is
+# 12,544 + ~8,800 = ~20.9 GiB used, ~3.1 GiB free — the band the 2026-10-02
+# shape hit, bought with precision instead of a slot. The 1024 ubatch
+# re-prices the compute buffers upward, so MEASURE after any change: if the
+# boot lands under 2 GiB free, step the pool down in 12,288-token
+# (384 MiB at this rate) increments.
 #
-# Sampling: greedy by default (the fleet contract); a client that sends a
-# temperature gets its own, and the harness applies per-model overrides
-# from models.yml.
+# KV arithmetic: attention KV exists only in the 8 full-attention layers
+# (4 KV heads x 256 head_dim x 2 B) = 16 KiB/token for K and the same again for
+# V => 32 KiB/token, so 12,288 tokens (= 6144/slot) is one 384 MiB step — that
+# is the ladder to walk if this moves again. Two native windows (2 x 262,144)
+# are NOT reachable with full-precision V: 524,288 x 32 KiB = 16.1 GiB of KV
+# against a card with ~15.7 GiB of KV room once weights, the vision tower, the
+# DeltaNet states and the graphs are loaded. 200,704 per stream is what full
+# precision buys; q8_0 on V would buy 262,144 instead.
 #
-# WARNING: do not name any variable TEMP, HOST, or GPU_MAX_HEAP_SIZE (HIP
-# reads those bare vars on the ROCm hosts -- see the 2026-08-13 incident in
-# AGENTS.md). This box is CUDA, but the rule keeps env files portable.
+# The asymmetric-KV contract this endpoint was created for (Frank, 2026-10-01
+# night: f16 K + q8_0 V) is now spent — it existed to make a third slot fit,
+# and the third slot is gone. The flags stay explicit so the choice is visible
+# in the argv either way.
+LLAMA_BIN=/home/dank/GitPackages/llama-master/build-cuda/bin/llama-server
+MODEL_FILE=/home/dank/models/Ornith-Q8/Ornith-1.5-9B-Q8_0.gguf
 
-# llama.cpp runtime binary: mainline master 37b53fd, CUDA sm_86,
-# FA_ALL_QUANTS=ON.
-LLAMA_BIN=/home/dank/GitPackages/llama.cpp-mainline/build-cuda/bin/llama-server
+# Vision: the Ornith Q8_0 GGUF ships no `v.*` tensors either (dense-9B byte
+# arithmetic: 9.79 GB ≈ 9B params x ~1.06 B/param leaves no room for a tower),
+# so the projector file IS the vision capability (mmproj-Ornith-1.5-9B-BF16
+# .gguf, 921,704,672 B, LFS oid 626f9f90..., fetched by
+# scripts/download-frank-small.sh). Empty string = text-only on purpose.
+MMPROJ_FILE=/home/dank/models/Ornith-Q8/mmproj-Ornith-1.5-9B-BF16.gguf
 
-# Model package. The MTP sidecar file is required for speculative decoding.
-MODEL_DIR=/home/dank/models/Gemma4-12B-QAT
-MODEL_FILE=Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced-Q4_K_M.gguf
-MMPROJ_PATH=/home/dank/models/Gemma4-12B-QAT/mmproj-Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced-BF16.gguf
-DRAFT_FILE=mtp-gemma-4-12B-it.gguf
-
-# Fleet identity and network. 7771 is this service's own endpoint on
-# Dankclaw; 7770 stays frank-large's. The alias is the id at /v1/models and
-# what clients ask for -- the harness entry in models.yml names it.
+# Fleet endpoint: same port and alias as every frank-small before it, so the
+# harness models.yml id and the homepage probe do not move.
 ALIAS=frank-small
 BIND_ADDR=0.0.0.0
 LISTEN_PORT=7771
 
-# Four FULL 262144-token contexts, one per slot (CTX_SIZE is the total
-# pool; --parallel divides it). See the SIZING note above for the fit
-# arithmetic; a fifth slot does not fit.
-CTX_SIZE=1048576
-PARALLEL=4
-CACHE_TYPE_K=q8_0
-CACHE_TYPE_V=q8_0
-# CPU RAM cache cap in MiB (build default 8192). This host is memory-tight
-# (frank-large's 70 GiB GTT plus the rest of the stack), and Gemma4 runs
-# fully on the 3090 (-ngl 999), so the CPU-side cache is not the hot path;
-# cap it at 2048 to give back ~6 GiB of LPDDR. Gemma4 is not a DeltaNet
-# model, so the >12.8 GiB CACHE_RAM floor does not apply.
-CACHE_RAM=2048
+# ONE conversation at the NATIVE 262,144 window (Frank, 2026-10-05). The
+# two-slot shape died under Ornith and the NO-QUANTIZE law, by arithmetic:
+# the KV rate is the ARCHITECTURE's, not the quant's — 9 full-attn layers x
+# 4 KV heads x 256 dim x 2 B = 36 KiB/token at f16 K+V (MiMo measured 24.5),
+# so 2 x 200,704 wants ~14.1 GiB of KV on top of +1,900 MiB of Q8 weights
+# against 2,279 MiB measured free; ubatch 1024 claws back 616, and no
+# pool step fits BOTH the 2-3 GiB floor and the harness compaction
+# invariant (slot >= 191,148). One native slot fits with room; the window
+# per conversation RISES 200,704 -> 262,144. The cost: a second concurrent
+# request queues behind the released slot (subagent concurrency 2 -> 1).
+CTX_SIZE=262144
+PARALLEL=1
 
-# Batch geometry: 8192 logical / 1024 physical. Measured on this card at
-# max length (254k-token prefill / 253k-ctx decode, 4 slots, q8 KV):
-# ubatch 512 -> pp 915 t/s, tg@250k 90-204; ubatch 1024 -> pp 962,
-# tg 257-302. Larger ubatches (2048) OOM the MTP draft context's pp
-# compute buffer (~3.3 GiB against 2.2 GiB headroom) and must not be
-# tried. Idle VRAM is 23,300 MiB at this geometry.
-BATCH_SIZE=8192
+# Compute-buffer geometry. This endpoint passed neither flag until 2026-10-03,
+# so it ran on llama.cpp's own defaults (2048/512); the pair is now 4096/2048.
+# Both are tunables because the pair is a measurement, not a law — and the
+# measurement is CARD-SPECIFIC. On the 2x R9700 layer-split sibling the ubatch
+# curve peaks at 512 and every step up costs GiB of compute buffer for no
+# speed (see frank-medium-batch-ubatch). On THIS CUDA 3090 the curve climbs
+# monotonically to 2048: 710 -> 1188 t/s prefill from ubatch 1024 to 2048 on
+# the 35B MoE sweep, 1944 -> 2423 t/s on the Ornith 8k sweep, with decode flat
+# to under 1 t/s on every rung. Ubatch is a prefill knob; decode never moves.
+#
+# The logical batch is free — two independent sweeps agree that at a fixed
+# ubatch, 2048/4096/8192 give identical prefill speed AND identical peak VRAM,
+# because the compute buffers are sized by the ubatch alone. Doubling it costs
+# nothing and buys nothing; it is here because it is the value the pair was
+# measured at.
+#
+# THE RISK IS ALL IN THE UBATCH. The 1024 -> 2048 step cost ~1,244 MiB of peak
+# graph VRAM on the 35B MoE (7,330 -> 8,574 MiB); this 9B hybrid has far
+# smaller compute tensors, so the delta should be well under that, but the
+# predecessor Q5_K_M shape recorded "ubatch 2048 failed the scratch boot" at
+# ~3,571 MiB free. The Q8_0 rung ships UBATCH 1024 (Frank, 2026-10-05: lower
+# ubatch FIRST — the measured 616 MiB step is the first VRAM the Q8 weights
+# claw back, before any pool move). Compute buffers are allocated at graph
+# capture during BOOT, so a failure is loud and fast, not a surprise at the
+# first long prompt. MEASURE nvidia-smi free immediately after the deploy:
+# under ~1.5 GiB, or a boot failure, step the POOL down next (12,288-token
+# increments, ~432 MiB each at the Ornith 36 KiB/token f16 KV rate).
+BATCH_SIZE=4096
 UBATCH_SIZE=1024
 
-# CPU threads: this service runs entirely on the GPU (NGL=999), so threads
-# cover tokenization, sampling and the HTTP layer only. Deliberately modest
-# -- this host's CPU is shared with halogen's engine, which is CPU-hungry.
-NUM_THREADS=8
-THREAD_BATCH=8
-
-# Every layer on the 3090, with flash attention (required for q8 KV).
-NGL=999
-FLASH_ATTN=on
-
-# Speculative decoding: the MTP sidecar (draft context) chained with the
-# draftless ngram-mod matcher. Draft depth and busy-wait level are made
-# explicit here (both were the implicit build defaults: 3 and 50).
-# Measured at 250k ctx: depth 4 + poll 100 DECODED WORSE (230-257 t/s)
-# than depth 3 + poll 50 (257-302 t/s) — the longer verification round
-# costs more attention re-read at long context than the extra accepted
-# token earns. Keep 3/50.
-SPEC_TYPE=draft-mtp,ngram-mod
-SPEC_DRAFT_N_MAX=3
-POLL_LEVEL=50
-
-# Greedy default (fleet contract); reasoning is extracted into
-# reasoning_content with an unlimited budget (the model decides).
-TEMPERATURE=0.0
-REASONING_FORMAT=deepseek
-REASONING_BUDGET=-1
+# The KV contract (2026-10-05, Frank): f16 K + f16 V — NO KV quantization.
+# The q8_0-V arithmetic was written and OVERRULED. The Q8 weights eat the
+# VRAM that funded f16 V on 2026-10-03, so the remedy ladder is UBATCH FIRST
+# (2048 -> 1024, the measured 616 MiB rung), then POOL STEPS — and
+# guess-and-check on the live card is the accepted method: boot, read
+# nvidia-smi, step down until the floor holds. --flash-attn stays pinned on
+# in the serve script: it is the fast path for f16 KV.
+CACHE_TYPE_K=f16
+CACHE_TYPE_V=f16
+NGL=99

scripts/frank-small.service

@@ -1,5 +1,5 @@
 [Unit]
-Description=llama.cpp Qwen3.8-27B GSQ-RCO IQ3_S MTP+vision (dankclaw, frank-small)
+Description=llama.cpp MiMo-V2.6-Distill-Qwen-9B Q6_K GGUF (qwen35 hybrid, 2x200704, unquantized f16 KV, mmproj vision) on the RTX 3090 (dankclaw, frank-small)
 After=network-online.target
 Wants=network-online.target
 
@@ -8,8 +8,7 @@ Type=simple
 User=dank
 Group=dank
 # CUDA pins its staging buffers, so the process needs the memlock budget even
-# though the weights themselves are copied into VRAM. Parity with the halogen
-# quadlet's Ulimit=memlock=-1:-1 and the retired ik unit.
+# though the weights themselves are copied into VRAM.
 LimitMEMLOCK=infinity
 Environment="FRANK_SMALL_ENV_FILE=/home/dank/serverAppConfigs/scripts/frank-small.env"
 ExecStart=/home/dank/serverAppConfigs/scripts/serve-frank-small.sh

scripts/patches/glm53-compile-01-model-decorator.patch

@@ -0,0 +1,86 @@
+# glm53-compile-01-model-decorator.patch — put the GLM-5.3-Flash trunk under torch.compile.
+#
+# STATUS: WIP — first of three patches in the "compile the GLM trunk" changelist. The
+# set does NOT yet boot the engine (see 03 for the next blocker), so it is NOT part of
+# the deployed frank-primary shape. Nothing in scripts/ mounts these files.
+#
+# WHY THIS IS THE RIGHT LEVER (measured 2026-09-30, not inferred):
+#
+# The fork's own flagship trunk — vllm/models/qwen4_exp/amd/model.py — carries
+# @support_torch_compile. vllm/models/glm5next/common/model.py does not, and the engine
+# says so on every boot:
+#
+#   Cudagraph mode FULL_AND_PIECEWISE requires piecewise capture, but the loaded model
+#   provides neither a compiled submodule nor breakable CUDA graphs.
+#   Overriding to FULL_DECODE_ONLY.
+#
+# A torch-profiler window on the live endpoint (496 decode tokens, rank0, 3 x 160-token
+# generations at the deployed shape) shows what the missing compilation costs. Total
+# device kernel time 46.18 s / 496 tokens = ~93 ms per token, against a measured step
+# time of ~104 ms (9.6 tok/s with the profiler attached):
+#
+#   16.07 s  34.8 %  708,500 launches  vectorized_elementwise_kernel      (22.7 us each)
+#    4.60 s  10.0 %    42,000           r4d_gemm_moe_w4a16_nt_b16_m64     (the MoE GEMM)
+#    3.11 s   6.7 %   174,708           elementwise_kernel_manual_unroll
+#    2.94 s   6.4 %   138,252           wvSplitK_hf_sml_                  (split-K GEMM)
+#    4.09 s   8.9 %    83,160           topk/router (sbtopk gatherTopK + warpMergeSortTopK
+#                                       + bitonicSortKVInPlace)
+#    3.01 s   6.5 %   133,920           r4d_mhc_* (hyper-connection pre/post/finish)
+#    2.78 s   6.0 %   127,916           inductor triton_poi_* / triton_per_* (the only
+#                                       fusions that DO happen)
+#    0.49 s   1.05%    20,664           r4d_moe_lru_gather_kernel         <-- the suspect
+#
+# That last row retired the fleet's working hypothesis. Decode on this shape is NOT
+# expert-gather bound: the gather is 1 % of device time, the same order as the telemetry
+# kernel that was switched on to measure it. 44 % of the budget is pointwise glue —
+# bf16<->fp32 conversions, adds, pows/rsqrts from fp32 RMSNorm, clamp/scale quant
+# passes — spread over ~1,428 launches per token. Inductor fusion is the only lever
+# that touches that mass; the router is second at 9 % but is binary-blocked (the
+# libr4d route kernel exists only as r4d_moe_route_kernel<128|256|512> instantiations
+# and GLM has 288 experts, with no HIP sources shipped to rebuild).
+#
+# TARGET: vllm/models/glm5next/common/model.py   (inside the pinned image)
+#   base    sha256 008eeaa2a07e091465319f86900b5a321d921238723ccbe22e6b9e971e1d2b28
+#   patched sha256 6e9cb81b84a9a88bd41403ca404aa04c3b791899dd77d661c3f25ae686cecd40
+#
+# APPLY: `patch -p1` from the vllm package root, or bind-mount new/.../model.py over
+# site-packages. Then R4D_GLM_COMPILE=1 (default) enables it; =0 makes the decorator a
+# no-op, so the file can stay mounted for an A/B without a second artifact.
+# Requires 02 (trace-safe probes) — without it the worker dies during profile_run on
+# the first probe call inside the traced region.
+--- a/vllm/models/glm5next/common/model.py
++++ b/vllm/models/glm5next/common/model.py
+@@ -1,12 +1,14 @@
+ # SPDX-License-Identifier: Apache-2.0
+ # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+ 
++import os
+ from collections.abc import Iterable
+ from typing import ClassVar, Literal
+ 
+ import torch
+ from torch import nn
+ 
++from vllm.compilation.decorators import support_torch_compile
+ from vllm.config import ParallelConfig, VllmConfig
+ from vllm.distributed import (
+     get_ep_group,
+@@ -623,6 +625,18 @@
+         )
+ 
+ 
++_GLM_COMPILE = os.environ.get("R4D_GLM_COMPILE", "1") == "1"
++
++
++@support_torch_compile(
++    dynamic_arg_dims={
++        "input_ids": 0,
++        "positions": -1,
++        "intermediate_tensors": 0,
++        "inputs_embeds": 0,
++    },
++    enable_if=lambda _cfg: _GLM_COMPILE,
++)
+ class Glm5NextModel(nn.Module):
+     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
+         super().__init__()

scripts/patches/glm53-compile-02-trace-safe-probes.patch

@@ -0,0 +1,119 @@
+# glm53-compile-02-trace-safe-probes.patch — libr4d capability probes must not do I/O
+# inside a torch.compile region.
+#
+# STATUS: WIP — second of three patches in the "compile the GLM trunk" changelist.
+# Inert (but harmless) without 01. Not part of the deployed shape.
+#
+# THE BREAK (measured: with 01 mounted and R4D_GLM_COMPILE=1, the worker dies during
+# profile_run, i.e. the first — cold — forward):
+#
+#   Explanation: Dynamo does not know how to trace the builtin `posix.stat`
+#   ... r4d_mhc_supported -> r4d_lib.r4d_mhc_enabled -> available -> os.path.exists
+#
+# THE NON-OBVIOUS PART: functools.lru_cache does NOT protect these probes. Dynamo
+# re-traces the wrapped body to build its guard even when the cache is warm — it warns
+# about exactly this ("Dynamo detected a call to a `functools.lru_cache`-wrapped
+# function ... Dynamo ignores the cache wrapper and directly traces the wrapped
+# function"). Verified in-image, torch 2.11.0+rocm10.0, torch.compile(backend="eager"):
+#
+#   cold lru_cache call       -> RAISE Unsupported
+#   WARM lru_cache call       -> RAISE Unsupported     <-- warming is not enough
+#   plain module-global read  -> compiles
+#   uncached os.path.exists   -> RAISE Unsupported
+#
+# So each probe must resolve ONCE into a plain value and be read as a global. The set
+# below is every no-argument probe / symbol lookup reachable from a forward pass,
+# grep-verified call sites:
+#   model_executor/layers/fused_moe/experts/r4d_w4a16_moe.py:79  available()
+#   model_executor/layers/fused_moe/experts/r4d_w4a16_moe.py:81  has_w4a16_moe()
+#   model_executor/kernels/mhc/r4d_dsv4.py:16                    r4d_mhc_enabled()
+#   v1/attention/ops/rocm_aiter_mla_sparse.py:694                r4d_indexer_enabled(), mqa_max_heads()
+#   v1/attention/ops/rocm_aiter_mla_sparse.py:1124               r4d_indexer_enabled()
+#   model_executor/layers/fused_moe/experts/r4d_mxfp4_moe.py:683 r4d_act_quant_enabled()
+#   model_executor/layers/fused_moe/experts/r4d_w4a16_moe.py _entry()  moe_w4a16()/moe_w4a16_mt1() per MoE call
+# (lib() is the transitive dependency of all of them.)
+#
+# TARGET: vllm/model_executor/kernels/r4d_lib.py   (inside the pinned image)
+#   base    sha256 b8dfeefb4acc72b67b0579a4ab0653ef5d48c40cc8649187d757c80ce6b9617a
+#   patched sha256 0751852e1ca813ad2bdb1da7bf10c605130e63e04ef9ec57e70b5f675a29c08e
+#
+# APPLY: append the block below to the module. R4D_FREEZE_PROBES=0 restores stock
+# behaviour. Cost when nothing is compiled: ten one-time probes at import (measured:
+# logged once per process, e.g. "libr4d capability probes frozen: 10 warm [...]").
+#
+# OBSERVED EFFECT: with 01+02 the worker gets past this break and reaches the next one
+# (see 03) — dynamo reports "Dynamo bytecode transform time: 1.08 s" for the trunk.
+--- a/vllm/model_executor/kernels/r4d_lib.py
++++ b/vllm/model_executor/kernels/r4d_lib.py
+@@ -361,3 +361,71 @@
+ @functools.lru_cache
+ def moe_w4a16_mt1():
+     return getattr(lib(), "r4d_gemm_moe_w4a16_nt_b16_m64_mt1", None)
++
++
++# FORK-LOCAL (patch: trace-safe capability probes).  torch.compile traces the
++# BODY of a functools.lru_cache call even when the cache is warm, so a capability
++# probe called from a forward pass is a dynamo graph break.  Measured 2026-09-30
++# on the GLM-5.3-Flash trunk under @support_torch_compile:
++#   Dynamo does not know how to trace the builtin `posix.stat`
++#   r4d_mhc_supported -> r4d_mhc_enabled -> available -> os.path.exists
++# Each probe below is wrapped to resolve ONCE into a memo and is warmed at import
++# (outside any tracing region), so a traced call is a dict read.  The probe set is
++# every no-argument capability probe / symbol lookup reachable from a forward
++# pass: the W4A16 + mxfp4 MoE dispatch, the mHC op, the MLA sparse indexer gate
++# and the draft/lm-head guards.
++_PROBE_NAMES = (
++    "available",
++    "lib",
++    "has_w4a16_moe",
++    "moe_w4a16",
++    "moe_w4a16_mt1",
++    "has_mhc_dsv4",
++    "r4d_mhc_enabled",
++    "r4d_indexer_enabled",
++    "mqa_max_heads",
++    "r4d_act_quant_enabled",
++)
++
++
++def _freeze_probe(fn):
++    """Return fn bound to its first result.  Result must be a plain value."""
++
++    memo = []
++
++    def probe(*args, **kwargs):
++        if not memo:
++            memo.append(fn(*args, **kwargs))
++        return memo[0]
++
++    return probe
++
++
++def _freeze_probes() -> list[str]:
++    frozen = []
++    for name in _PROBE_NAMES:
++        fn = globals().get(name)
++        if fn is None:
++            continue
++        # Only the fast branch of a bound probe is traced; an empty memo means the
++        # first call still does I/O, so populate it here as well.
++        globals()[name] = _freeze_probe(fn)
++        frozen.append(name)
++    return frozen
++
++
++from vllm.logger import init_logger as _init_logger
++
++_probe_log = _init_logger(__name__)
++
++if os.environ.get("R4D_FREEZE_PROBES", "1") == "1":
++    _warmed, _failed = [], []
++    for _n in _freeze_probes():
++        try:
++            globals()[_n]()
++            _warmed.append(_n)
++        except Exception:  # noqa: BLE001 -- probe unavailable here; recompute lazily
++            _failed.append(_n)
++    _probe_log.info("libr4d capability probes frozen: %d warm %s%s",
++                    len(_warmed), _warmed,
++                    f" (deferred: {_failed})" if _failed else "")

scripts/patches/glm53-compile-03-fullgraph-optout.patch

@@ -0,0 +1,116 @@
+# glm53-compile-03-fullgraph-optout.patch — let torch.compile split the graph instead
+# of failing on a break.
+#
+# STATUS: WIP — third of three patches in the "compile the GLM trunk" changelist.
+# With 01+02+03 the model TRACES, then hits the NEXT blocker (below). The set does not
+# yet boot the engine; nothing in scripts/ mounts it.
+#
+# THE BREAK: with 01+02 mounted the worker dies with
+#
+#   RuntimeError: Graph breaks are not supported with aot compile.
+#   Please use torch.compile(fullgraph=True).
+#
+# because vllm/compilation/wrapper.py calls torch.compile(..., fullgraph=True, ...) and
+# this build has AOT compile on by default (VLLM_USE_AOT_COMPILE). The GLM trunk cannot
+# satisfy fullgraph: its indexer metadata path syncs to the host —
+#   models/glm5next/amd/sparse_indexer.py:263  int(positions[...].max().item()) + 1
+#   models/glm5next/amd/sparse_indexer.py:419  lmax = int(decode_metadata.decode_lens.max().item())
+# — and a .item() inside a captured graph is not expressible.
+#
+# WHAT WAS VERIFIED (in-image, torch 2.11.0+rocm10.0, torch.compile(backend="eager")):
+#
+#   fullgraph=True   suppress_errors=False  -> RAISE Unsupported
+#   fullgraph=True   suppress_errors=True   -> RAISE Unsupported   <-- suppress_errors
+#                                                                     does NOT cover breaks
+#   fullgraph=False  suppress_errors=False  -> ok   (traceable segments compiled by
+#                                                    inductor, the rest runs eagerly;
+#                                                    numerics unchanged)
+#
+# So breaks must be TOLERATED, not suppressed: the traceable mass still fuses. This
+# patch makes the fullgraph requirement opt-out via VLLM_COMPILE_FULLGRAPH=0 (default 1,
+# so every other model keeps stock behaviour), and the run also needs
+# VLLM_USE_AOT_COMPILE=0 because AOT compile refuses breaks outright.
+#
+# TARGET: vllm/compilation/wrapper.py   (inside the pinned image)
+#   base    sha256 969961dca41c78385212e985cfc0c57ef067f4a2d7490c73c658c600a1b95cac
+#   patched sha256 732363bb1a1367772f13c60a655c678e7e2e9e56186677b14238f5e84cac31b3
+#   (the file already imports os; the patch does not add a second import)
+#
+# NEXT BLOCKER — NOT FIXED HERE. With 01+02+03 and
+# R4D_GLM_COMPILE=1 VLLM_COMPILE_FULLGRAPH=0 VLLM_USE_AOT_COMPILE=0 the trunk traces
+# ("Dynamo bytecode transform time: 1.08 s") and then the worker dies during profile_run
+# on the decode forward:
+#
+#   KeyError: 'in_proj_qkvbfg_a'
+#     model.py:748  Glm5NextModel.forward        hidden_states, residual, post, comb = layer(...)
+#     model.py:456  Glm5NextDecoderLayer.forward
+#
+# `in_proj_qkvbfg_a` is the KDA layer's merged column-parallel projection
+# (models/glm5next/common/kda.py:217, called at :407) — and it is also a name in the
+# load_weights shard table (model.py:780-785), i.e. a real submodule attribute name,
+# not an arbitrary string.
+#
+# WHERE IT IS THROWN (established 2026-09-30 by reading the full traceback, since the
+# report line is inside dynamo-rewritten code): torch/_dynamo keeps a name -> module map
+# and looks it up by name — output_graph.py:1259 / :1384 register into `nn_modules`,
+# :3000 does `self.nn_modules[node.target]` (a bare-name KeyError when absent), and
+# :1225 documents stripping "_buffers[..]/_parameters[..]/_modules[..]" from those
+# names. So the compiler is failing to RESOLVE A SUBMODULE BY NAME while tracing the
+# GLM decoder layer. Whether the root cause is the fork's module tree (something
+# dynamo cannot register statically) or dynamo's own registration is NOT yet settled.
+#
+# STATUS OF THAT BLOCKER (updated 2026-09-30, after the diagnostics below):
+#   * TORCHDYNAMO_VERBOSE=1 adds NO internal stack trace -> it is a plain Python
+#     KeyError raised in a callee dynamo INLINED into Glm5NextDecoderLayer.forward
+#     (the reported line 456 is that method's `def` line, which is how inlined frames
+#     surface).
+#   * It is NOT raised from dynamo's four nn_modules lookups: the probe in
+#     /home/dank/dynamo-probe wraps three of them and stayed silent over two runs with
+#     the probe confirmed installed in the API-server and worker processes.
+#   * It is NOT a literal string lookup: the name appears tree-wide only as the KDA
+#     layer's attribute (kda.py:217, used at :407) and in the load_weights shard table
+#     (model.py:780-785). The four `proj_name = prefix.split(".")[-1]` sites in the
+#     quantization utils do compute exactly this bare key, but each is guarded
+#     (`if proj_name not in fused_mapping: return`), so none of them can raise it.
+#   * It reproduces on every run of this configuration (5 of 5).
+# REMAINING INSTRUMENT: name the inlined caller. Either `TORCH_LOGS=+dynamo` on a boot
+# and grep the log for the key (dynamo prints inlined frames with source), or wrap the
+# inlining candidates in that forward — the mHC CustomOps (hc_pre / hc_post /
+# hc_fused_post_pre in layers/mhc.py) and the indexer op — with a KeyError-reporting
+# probe. Until that names the caller this is a dynamo-side name-resolution failure.
+# NOTE the launcher gotcha that wasted a boot earlier: the scratch launcher now
+# preserves ENVARGS (`ENVARGS="${ENVARGS:-}"`), but it previously reset it, silently
+# dropping outer env vars.
+#
+# SECOND BLOCKER, KNOWN AND NOT FATAL IN THIS CONFIG: the indexer's in-forward
+# workspace acquisition (patch 05 makes it compiler-opaque) and the fused-norm
+# launcher cell (patch 04 removes the call) are real breaks, but with
+# VLLM_COMPILE_FULLGRAPH=0 they are tolerated — they only become fatal under
+# fullgraph=True, which this trunk can never satisfy anyway.
+--- a/vllm/compilation/wrapper.py
++++ b/vllm/compilation/wrapper.py
+@@ -19,6 +19,15 @@
+ 
+ logger = init_logger(__name__)
+ 
++
++# FORK-LOCAL (patch: fullgraph opt-out).  torch.compile(fullgraph=True) turns any
++# dynamo graph break into a hard failure; suppress_errors does NOT cover breaks
++# (verified in-image, torch 2.11+rocm10.0).  With VLLM_COMPILE_FULLGRAPH=0 the
++# traceable segments are still compiled by inductor and the untraceable ones run
++# eagerly, which is what a model with device-sync metadata (the GLM-5.3-Flash
++# indexer) needs.  Default 1 keeps stock behaviour for every other model.
++_FULLGRAPH = os.environ.get("VLLM_COMPILE_FULLGRAPH", "1") == "1"
++
+ R = TypeVar("R")
+ P = ParamSpec("P")
+ 
+@@ -146,7 +155,7 @@
+         with aot_context:
+             self._compiled_callable = torch.compile(
+                 compiled_ptr,
+-                fullgraph=True,
++                fullgraph=_FULLGRAPH,
+                 dynamic=False,
+                 backend=backend,
+                 options=options,

scripts/patches/glm53-compile-04-fused-qkv-rmsnorm-optout.patch

@@ -0,0 +1,112 @@
+# glm53-compile-04-fused-qkv-rmsnorm-optout.patch — take GLM's fused-norm JIT kernel
+# out of the traced path.
+#
+# STATUS: WIP — fourth of the "compile the GLM trunk" changelist. Needs 01 (the
+# decorator) to matter; not part of the deployed shape.
+#
+# MEASUREMENT NOTE — read the run history before trusting any claim below, because a
+# launcher bug made four consecutive boots misleading.  The scratch launcher
+# (serve_glm_autoround.sh) was restructured on 2026-09-30 and the
+# `-e VLLM_COMPILE_FULLGRAPH=0` line was DROPPED, so those boots mounted the wrapper
+# patch while `_FULLGRAPH` still defaulted to True.  With fullgraph=True *every* break
+# is an error, which is why the reported failure kept moving as breaks were removed:
+#
+#   fullgraph=False, AOT off  -> breaks tolerated; dies later on the dynamo name-map
+#                                `KeyError: 'in_proj_qkvbfg_a'`
+#   fullgraph=True,  AOT off  -> dies on whichever break comes first
+#                                (cell -> workspace -> compiler-disable)
+#
+# An earlier version of this note blamed `TORCH_LOGS=graph_breaks` for the difference.
+# That was WRONG — the flag only logs; the launcher silently ran fullgraph=True.
+# Launcher fixed; the env var is set again.
+#
+# What it means for THIS patch: under fullgraph=True the fused-norm call was the first
+# break dynamo reported, and after this patch the report moved on to the workspace
+# acquisition — so it removes a real break.  It is not by itself what unblocks a boot:
+# with breaks genuinely tolerated the blocker is the name-map KeyError.
+#
+# IT IS STILL THE RIGHT PATTERN TO REMOVE, either way: the launcher family
+# (@kernel_launcher, 39 call sites) is exactly the shape that produces breaks, and the
+# object this one closes over is bound after use (see below).
+#
+# THE BREAK (measured, TORCH_LOGS=graph_breaks, rank0, 2026-09-30).  With 01+02+03 the
+# trunk traces, and with AOT off + fullgraph=True dynamo reports a FATAL break — fatal
+# even with VLLM_COMPILE_FULLGRAPH=0, because dynamo cannot build a resume function
+# around it:
+#
+#   torch._dynamo.exc.Unsupported: Read uninitialized cell
+#   Explanation: Attempted to read a cell variable that has not been populated yet.
+#   Developer debug context: CellVariable()
+#
+# User stack:
+#   glm5next/common/model.py:748    Glm5NextModel.forward       layer(...)
+#   glm5next/common/model.py:526    Glm5NextDecoderLayer.forward
+#   glm5next/common/attention.py:590
+#   model_executor/layers/mla.py:208
+#     q_proj_input, kv_c_normed = fused_q_kv_rmsnorm(...)
+#   -> torch/_dynamo/polyfills/__init__.py:392 getattr_and_trace
+#
+# So dynamo is tracing the fork's JIT-kernel launcher framework
+# (VllmTritonJitKernel + @kernel_launcher — 39 call sites) and finds a closure name
+# that is not yet bound.  In models/common/ops/fused_qk_rmsnorm.py the kernel object
+# is bound AFTER the function that closes over it:
+#
+#   def fused_q_kv_rmsnorm(...):          # line 194
+#       ...
+#       _FUSED_Q_KV_RMSNORM_KERNEL(...)   # line 211
+#   _FUSED_Q_KV_RMSNORM_KERNEL = FusedQKVRMSNormKernel()    # line 223
+#
+# Two ways out, and this patch takes the cheap one:
+#   (a) WORKAROUND (here): stop calling the fused kernel from the traced region.  The
+#       flag is already a constructor argument (mla.py:80, default False) that glm5next
+#       turns ON at attention.py:583, so it becomes an env knob.
+#   (b) ROOT FIX (not done): make the launcher trace-safe for all 39 sites.  Any other
+#       @kernel_launcher-backed kernel still in the traced path can produce the same
+#       fatal break, so (b) is the durable fix and (a) is what unblocks iterating.
+#
+# COST of R4D_GLM_FUSE_QKV_RMSNORM=0: one extra small RMSNorm launch per MLA layer per
+# token (q_a_layernorm + kv_a_layernorm instead of one fused kernel) — ~2 launches out
+# of ~3,700 per token on this shape, against the 44 % of device time currently spent in
+# unfused pointwise work.  Correct trade for a compile experiment.
+#
+# TARGET: vllm/models/glm5next/common/attention.py   (inside the pinned image)
+#   base    sha256 362bbf0e6c50cdfa446c6551b9b1ddff425e08f88e3d4140fd8db88f8a619e0c
+#   patched sha256 84969993c250e9f8183a33a6e1ddd607272f48b950b560272eb33b7cdde31189
+#
+# APPLY: `patch -p1` from the vllm package root (verified: applies clean, reproduces
+# the patched sha256), or bind-mount new/.../attention.py over site-packages.
+# R4D_GLM_FUSE_QKV_RMSNORM=1 (default) restores the fused kernel.
+--- a/vllm/models/glm5next/common/attention.py
++++ b/vllm/models/glm5next/common/attention.py
+@@ -1,6 +1,7 @@
+ # SPDX-License-Identifier: Apache-2.0
+ # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+ 
++import os
+ import torch
+ import torch.nn.functional as F
+ from torch import nn
+@@ -36,6 +37,14 @@
+ from vllm.utils.deep_gemm import PAGED_MQA_PAGE_SIZES
+ from vllm.v1.kv_cache_interface import KpoolTailSpec, MLAAttentionSpec
+ 
++
++# FORK-LOCAL (patch: keep the traced region trace-safe).  R4D_GLM_FUSE_QKV_RMSNORM=0
++# replaces the fused `fused_q_kv_rmsnorm` JIT-kernel launch (one kernel) with two plain
++# RMSNorms, because dynamo cannot trace the launcher's closure ("Read uninitialized cell",
++# fatal even with fullgraph=False).  The fused kernel object in
++# models/common/ops/fused_qk_rmsnorm.py is bound AFTER the function that uses it.
++_FUSE_QKV_RMSNORM = os.environ.get("R4D_GLM_FUSE_QKV_RMSNORM", "1") == "1"
++
+ logger = init_logger(__name__)
+ 
+ # Shared torch.compile config for the indexer's small-kernel leaves. The MLA
+@@ -580,7 +589,7 @@
+             quant_config,
+             prefix,
+             skip_topk=False,
+-            fuse_qkv_rmsnorm=True,
++            fuse_qkv_rmsnorm=_FUSE_QKV_RMSNORM,
+         )
+ 
+     def forward(

scripts/patches/glm53-compile-05-workspace-acquisition-opaque.patch

@@ -0,0 +1,96 @@
+# glm53-compile-05-workspace-acquisition-opaque.patch — the indexer's workspace
+# acquisition must not be traced.
+#
+# STATUS: WIP — fifth of the "compile the GLM trunk" changelist. This is a break worth
+# removing, but NOT part of the confirmed blocker and NOT required for the
+# break-tolerant route: `torch.compiler.disable` is itself a graph break, so this patch
+# helps only a hypothetical fullgraph=True route, where the workspace acquisition would
+# be better RESTRUCTURED out of the forward (allocate at init/profile time) than made
+# opaque. Kept because the discovery is real and the wrapper is a no-op when nothing is
+# compiling.
+#
+# THE BREAK — CONFIRMED with AND WITHOUT the logging flag (2026-09-30):
+#
+#   torch._dynamo.exc.Unsupported: Unsupported method call
+#   user stack
+#     glm5next/common/model.py:748      Glm5NextModel.forward
+#     glm5next/common/model.py:526      Glm5NextDecoderLayer.forward
+#     glm5next/common/attention.py:599 / :401
+#     glm5next/amd/sparse_indexer.py:711  forward_native
+#     glm5next/amd/sparse_indexer.py:132  sparse_attn_indexer_kpool
+#     v1/worker/workspace.py:135          get_simultaneous
+#     v1/worker/workspace.py:155          _ensure_workspace_size
+#
+# Line 132 is the PROFILE-RUN branch — the source comment says what it is:
+#
+#     if not isinstance(attn_metadata, dict):
+#         # Reserve workspace for indexer during profiling run
+#
+# so the engine dies exactly where it sizes its memory.  `_ensure_workspace_size`
+# reads ContextVar and ubatch state (`_workspace_lane.get()`,
+# `dbo_current_ubatch_id()`) and may allocate; none of that is traceable.
+#
+# FIX: the documented escape hatch — `torch.compiler.disable` around a thin wrapper,
+# so dynamo stops analysing the call and executes it eagerly from inside the compiled
+# region.  Chosen over decorating the workspace manager's own methods because that
+# would change workspace behaviour for every model, not just this one.  The wrapper is
+# a no-op when nothing is compiling, so it needs no env gate and is safe in the
+# stock (uncompiled) path — which is why this patch can be reasoned about
+# independently of whether the changelist ever ships.
+#
+# CAVEAT, stated plainly: a compiler-disabled region is still a graph break, so this
+# needs patch 03 (VLLM_COMPILE_FULLGRAPH=0) to be tolerable at all, and whether dynamo
+# can build a resume function around it is exactly what the next boot decides — the
+# difference between this failure and 04's `Read uninitialized cell`.
+#
+# TARGET: vllm/models/glm5next/amd/sparse_indexer.py   (inside the pinned image)
+#   base    sha256 f773a99265d3f440c8fad22c8fb5a8b33f926162094cffc3a2ecdaa42763c5b1
+#   patched sha256 c154115afe8253a35c652b7198929ef4c1874957d333fdb3340cacf88e612b78
+#
+# APPLY: `patch -p1` from the vllm package root, or bind-mount over site-packages.
+--- a/vllm/models/glm5next/amd/sparse_indexer.py
++++ b/vllm/models/glm5next/amd/sparse_indexer.py
+@@ -35,6 +35,21 @@
+ from vllm.v1.attention.ops.common import pack_seq_triton, unpack_seq_triton
+ from vllm.v1.worker.workspace import current_workspace_manager
+ 
++@torch.compiler.disable
++def _acquire_workspace(*shapes_and_dtypes):
++    """Acquire shared workspace buffers, untraced.
++
++    Workspace acquisition reads ContextVar/ubatch state and may allocate
++    (v1/worker/workspace.py: ``_ensure_workspace_size`` -> ``_workspace_lane.get()``,
++    ``dbo_current_ubatch_id()``), which dynamo cannot trace -- it raises
++    ``Unsupported: Unsupported method call`` and kills the engine during the profile
++    run.  ``torch.compiler.disable`` keeps the call as an opaque, eagerly-executed
++    region inside the compiled graph.  No-op when not compiling.
++    """
++
++    return current_workspace_manager().get_simultaneous(*shapes_and_dtypes)
++
++
+ logger = init_logger(__name__)
+ 
+ # kpool write helper: form pools from the current token batch and compress them
+@@ -129,7 +144,7 @@
+         values_spec, scales_spec = _gather_workspace_shapes(
+             total_seq_lens, head_dim, fp8_dtype, use_fp4_cache
+         )
+-        current_workspace_manager().get_simultaneous(
++        _acquire_workspace(
+             values_spec,
+             scales_spec,
+             ((RADIX_TOPK_WORKSPACE_SIZE,), torch.uint8),
+@@ -275,11 +290,10 @@
+         # Layout switches between FP8 (head_dim bytes + 4-byte fp32 scale) and
+         # MXFP4 (head_dim/2 bytes packed + head_dim/MXFP4_BLOCK_SIZE ue8m0
+         # scales) based on use_fp4_cache.
+-        workspace_manager = current_workspace_manager()
+         values_spec, scales_spec = _gather_workspace_shapes(
+             total_seq_lens, head_dim, fp8_dtype, use_fp4_cache
+         )
+-        k_quant_full, k_scale_full = workspace_manager.get_simultaneous(
++        k_quant_full, k_scale_full = _acquire_workspace(
+             values_spec,
+             scales_spec,
+         )

scripts/patches/glm53-dflash-01-aux-hidden-states.patch

@@ -0,0 +1,173 @@
+# glm53-dflash-01-aux-hidden-states.patch — give the GLM trunk the EAGLE3 / DFlash2
+# auxiliary-hidden-state interface.
+#
+# STATUS: WIP, and INDEPENDENT of the compile changelist (01-05). It patches the same
+# file as glm53-compile-01-model-decorator.patch, so if both are ever applied, apply 01
+# first. Not part of the deployed shape.
+#
+# WHY.  `--speculative-config '{"method":"dflash","model":"/path/to/drafter", ...}'` with
+# incoai/GLM-5.3-Flash-DFlash2 is accepted -- the config parses, the drafter loads -- and
+# then the worker dies at init with:
+#
+#     RuntimeError: Model does not support EAGLE3 interface
+#
+# raised by v1/worker/gpu/spec_decode/eagle/eagle3_utils.py:20 through
+# `supports_eagle3(model)` == `isinstance(model, SupportsEagle3)` (a runtime-checkable
+# Protocol).  Nothing about the drafter, the method name or the payload is wrong:
+# `glm5next` implements neither half of the contract.
+#
+# THE CONTRACT (read out of this image, not assumed):
+#   * DEFINITIVE (v5, after four failed boots): the member set is checked against
+#     the INSTANCE the runner calls -- which for GLM-5.3-Flash is
+#     Glm5NextForConditionalGeneration, NOT Glm5NextForCausalLM.  The multimodal
+#     wrapper wraps the text model as self.language_model
+#     (init_vllm_registered_model, architectures=["Glm5NextForCausalLM"],
+#     glm5next/common/model.py:1113), and model_runner.py:425 calls
+#     set_eagle3_aux_hidden_state_layers(self.model) on the CONDITIONAL.
+#     Presence rule (verified in-image with stock typing, python 3.14):
+#     isinstance walks ALL __protocol_attrs__ -- members of SupportsEagle3 AND
+#     its base SupportsEagleBase (has_own_lm_head, has_own_embed_tokens) -- and
+#     getattr_static()s each one off the instance, breaking on the first miss.
+#     So BOTH wrappers need all three attributes.
+#   * MRO law: INHERITING SupportsEagle3 as a base dies at class creation with
+#     "Cannot create a consistent method resolution order (MRO) for bases ..."
+#     (GLM's wrappers already carry four/five Protocol-derived bases).  The
+#     attribute route avoids the MRO entirely.  Also proven by simulation:
+#     the four-base ForCausalLM ordering cannot take SupportsEagle3 in ANY
+#     position.
+#   * Method placement: without inheriting, the Protocol's default
+#     set_aux_hidden_state_layers / get_eagle3_default_aux_hidden_state_layers
+#     do not exist on either wrapper, so both are defined explicitly --
+#     ForCausalLM unwraps to self.model (the EagleModelMixin trunk, v1's
+#     correct bodies), the conditional delegates to self.language_model
+#     (the pixtral.py:498 pattern).
+#   * interfaces.py:1673-1719 -- for reference, the inherited defaults unwrap
+#     `holder = getattr(parent_ref, "model", parent_ref)` and assert
+#     `isinstance(holder, EagleModelMixin)`, defaulting to (2, L // 2, L - 3).
+#   * eagle3_utils.py:26-31 -- layers come from the spec config when it carries them
+#     (this drafter ships target_layer_ids [5, 14, 24, 33, 42]), else from the default.
+#   * deepseek_v2.py:1551-1585 and v1/worker/gpu/cudagraph_utils.py:621 -- the trunk
+#     appends one tensor per aux layer while looping and returns the TUPLE
+#     (hidden_states, aux_hidden_states) only when that list is non-empty.
+#
+# WHAT IT ADDS for GLM: EagleModelMixin on Glm5NextModel; aux capture in the layer
+# loop, where the aux state is the single-stream `hidden_states` ENTERING that layer
+# (GLM's loop carries the already-summed single-stream state -- its non-mHC branch
+# re-splits `residual = hidden_states` inside the layer -- so this is the equivalent
+# of deepseek_v2's `hidden_states + residual`); the conditional tuple return; the
+# three SupportsEagle3 member attributes on BOTH wrappers; and the two interface
+# methods on both (ForCausalLM: unwrap to trunk; conditional: delegate to
+# language_model).
+#
+# HONEST SCOPE: this makes the interface PRESENT, which is exactly what the engine
+# checks -- it does not make drafts GOOD, and per the measured profile it is not a
+# decode-speed fix (a verification pass pays pointwise cost proportional to its token
+# count, which is why MTP4 measured flat).  It is the prerequisite for driving DFlash2
+# on this endpoint at all.
+#
+# TARGET: vllm/models/glm5next/common/model.py   (inside the pinned image)
+#   base    sha256 008eeaa2a07e091465319f86900b5a321d921238723ccbe22e6b9e971e1d2b28
+#   patched sha256 bc9913a15c8e7f88e0c36b07bf029b383cc8c2078b58d66dc91aa235f8af77f4
+#
+# APPLY: `patch -p1` from the vllm package root, or bind-mount the patched file over
+# site-packages (which is what a scratch-launcher DFLASH_PATCH=1 does).
+--- a/vllm/models/glm5next/common/model.py
++++ b/vllm/models/glm5next/common/model.py
+@@ -95,6 +95,12 @@
+ logger = init_logger(__name__)
+ 
+ 
++from vllm.model_executor.models.interfaces import (
++    EagleModelMixin,
++    SupportsEagle3,
++)
++
++
+ class Glm5NextMLP(nn.Module):
+     def __init__(
+         self,
+@@ -623,7 +629,7 @@
+         )
+ 
+ 
+-class Glm5NextModel(nn.Module):
++class Glm5NextModel(nn.Module, EagleModelMixin):
+     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
+         super().__init__()
+ 
+@@ -730,7 +736,14 @@
+         if self.is_sequence_parallel:
+             hidden_states = sp_shard(hidden_states)
+ 
+-        for layer in self._active_layers:
++        aux_hidden_states: list[torch.Tensor] = []
++        for local_idx, layer in enumerate(self._active_layers):
++            if (getattr(self, "start_layer", 0) + local_idx) in self.aux_hidden_state_layers:
++                # Auxiliary hidden state for EAGLE-3 / DFlash2 drafts: the
++                # single-stream state ENTERING this layer, before its input norm --
++                # the same semantic deepseek_v2 implements by appending
++                # hidden_states + residual at this point in its loop.
++                aux_hidden_states.append(hidden_states)
+             hidden_states, residual, post, comb = layer(
+                 positions, hidden_states, residual, post, comb
+             )
+@@ -749,6 +762,8 @@
+             hidden_states = sp_all_gather(hidden_states)[:full_num_tokens]
+ 
+         hidden_states = self.norm(hidden_states)
++        if len(aux_hidden_states) > 0:
++            return hidden_states, aux_hidden_states
+         return hidden_states
+ 
+     def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
+@@ -937,6 +952,10 @@
+ class Glm5NextForCausalLM(
+     nn.Module, HasInnerState, SupportsPP, MixtureOfExperts, IsHybrid
+ ):
++    supports_eagle3: ClassVar[Literal[True]] = True
++    has_own_lm_head: bool = False
++    has_own_embed_tokens: bool = False
++
+     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
+         super().__init__()
+         self.model_config = vllm_config.model_config
+@@ -963,6 +982,14 @@
+     def embed_input_ids(self, input_ids: torch.Tensor) -> torch.Tensor:
+         return self.model.embed_input_ids(input_ids)
+ 
++    def set_aux_hidden_state_layers(self, layers: tuple[int, ...]) -> None:
++        # EAGLE-3 / DFlash2 interface: unwrap to the trunk (the EagleModelMixin).
++        self.model.aux_hidden_state_layers = layers
++
++    def get_eagle3_default_aux_hidden_state_layers(self) -> tuple[int, ...]:
++        num_layers = len(self.model.layers)
++        return (2, num_layers // 2, num_layers - 3)
++
+     def forward(
+         self,
+         input_ids: torch.Tensor | None,
+@@ -1033,6 +1060,10 @@
+ class Glm5NextForConditionalGeneration(
+     Glm4vForConditionalGeneration, HasInnerState, IsHybrid, MixtureOfExperts
+ ):
++    supports_eagle3: ClassVar[Literal[True]] = True
++    has_own_lm_head: bool = False
++    has_own_embed_tokens: bool = False
++
+     # The text model (KDA + dense-MLA + MoE) is a hybrid mamba model. The
+     # multimodal wrapper must declare the same interfaces so vLLM treats it as
+     # hybrid (auto-aligns mamba/attention block sizes, sizes the mamba state
+@@ -1076,6 +1107,13 @@
+ 
+         return Glm5NextForCausalLM.get_mamba_state_copy_func()
+ 
++    def set_aux_hidden_state_layers(self, layers: tuple[int, ...]) -> None:
++        # EAGLE-3 / DFlash2: delegate to the language model (pixtral.py:498).
++        self.language_model.set_aux_hidden_state_layers(layers)
++
++    def get_eagle3_default_aux_hidden_state_layers(self) -> tuple[int, ...]:
++        return self.language_model.get_eagle3_default_aux_hidden_state_layers()
++
+     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
+         super(Glm4vForConditionalGeneration, self).__init__()
+         config = vllm_config.model_config.hf_config

scripts/patches/glm53-dflash-04-kv-groups-draft.patch

@@ -0,0 +1,317 @@
+--- a/vllm/v1/core/kv_cache_utils.py
++++ b/vllm/v1/core/kv_cache_utils.py
+@@ -1210,17 +1210,38 @@
+         for name, spec in kv_cache_spec.items()
+         if isinstance(spec, KpoolTailSpec)
+     }
++    hidden_specs = {
++        name: spec
++        for name, spec in kv_cache_spec.items()
++        if isinstance(spec, HiddenStateCacheSpec)
++    }
+     attn_specs = {
+         name: spec
+         for name, spec in kv_cache_spec.items()
+-        if not isinstance(spec, (MambaSpec, KpoolTailSpec))
++        if not isinstance(
++            spec, (MambaSpec, KpoolTailSpec, HiddenStateCacheSpec)
++        )
+     }
+-    if not mamba_specs or not all(
+-        type(spec) is MLAAttentionSpec for spec in attn_specs.values()
+-    ):
++    if not mamba_specs or not attn_specs:
++        return None
++    # FORK-LOCAL (patches/glm53-dflash-04-kv-groups-draft): with a dflash
++    # drafter, attn_specs also carries the drafter's attention layers. The
++    # GLM MLA layers keep their own group (no page unification -- per-group
++    # pages are the glm5 design); the drafter layers must form one uniform
++    # group of their own, and the hidden-state layers get per-layer groups
++    # like the generic tail.
++    mla_specs = {
++        name: spec
++        for name, spec in attn_specs.items()
++        if type(spec) is MLAAttentionSpec
++    }
++    draft_specs = {
++        name: spec for name, spec in attn_specs.items() if name not in mla_specs
++    }
++    if not mla_specs:
++        return None
++    if draft_specs and UniformTypeKVCacheSpecs.from_specs(draft_specs) is None:
+         return None
+-
+-    mla_specs = cast(dict[str, MLAAttentionSpec], attn_specs)
+     idx_pages = {
+         spec.page_size_bytes for spec in mla_specs.values() if spec.tokens_per_state > 1
+     }
+@@ -1233,7 +1254,7 @@
+     mla_pages = {mla_specs[name].page_size_bytes for name in mla_names}
+     assert len(mla_pages) == 1
+     mla_page = mla_pages.pop()
+-    uniform_spec = UniformTypeKVCacheSpecs.from_specs(attn_specs)
++    uniform_spec = UniformTypeKVCacheSpecs.from_specs(mla_specs)
+     assert uniform_spec is not None
+ 
+     tail_group: KVCacheGroupSpec | None = None
+@@ -1270,10 +1291,20 @@
+     for index, name in enumerate(mamba_specs):
+         mamba_grouped_names[index % num_groups].append(name)
+ 
++    draft_group: KVCacheGroupSpec | None = None
++    if draft_specs:
++        draft_uniform = UniformTypeKVCacheSpecs.from_specs(draft_specs)
++        assert draft_uniform is not None
++        draft_group = KVCacheGroupSpec(list(draft_specs), draft_uniform)
++    hidden_groups = [
++        KVCacheGroupSpec([name], spec) for name, spec in hidden_specs.items()
++    ]
+     return (
+-        [KVCacheGroupSpec(list(attn_specs), uniform_spec)]
++        [KVCacheGroupSpec(list(mla_specs), uniform_spec)]
+         + ([tail_group] if tail_group is not None else [])
++        + ([draft_group] if draft_group is not None else [])
+         + create_kv_cache_group_specs(padded_specs, mamba_grouped_names)
++        + hidden_groups
+     )
+ 
+ 
+@@ -1289,10 +1320,17 @@
+         int,
+         list[str],
+         int,
++        KVCacheGroupSpec | None,
++        list[KVCacheGroupSpec],
+     ]
+     | None
+ ):
+-    """Recognize the GLM-5.3-Flash grouping after optional PP projection."""
++    """Recognize the GLM-5.3-Flash grouping after optional PP projection.
++
++    FORK-LOCAL (patches/glm53-dflash-04-kv-groups-draft): also recognizes the
++    drafter's uniform group and the per-layer hidden-state groups, returned
++    as ``draft_group`` / ``hidden_groups`` so the pool builder and the memory
++    estimator can size their tensors beside the aliased GLM pool."""
+     uniform_groups = [
+         group
+         for group in kv_cache_groups
+@@ -1301,17 +1339,28 @@
+     mamba_groups = [
+         group for group in kv_cache_groups if isinstance(group.kv_cache_spec, MambaSpec)
+     ]
++    hidden_groups = [
++        group
++        for group in kv_cache_groups
++        if isinstance(group.kv_cache_spec, HiddenStateCacheSpec)
++    ]
+     attn_group: KVCacheGroupSpec | None = None
+     tail_group: KVCacheGroupSpec | None = None
++    draft_group: KVCacheGroupSpec | None = None
+     for group in uniform_groups:
+         inner = cast(UniformTypeKVCacheSpecs, group.kv_cache_spec).kv_cache_specs
+         if all(type(spec) is MLAAttentionSpec for spec in inner.values()):
+             attn_group = group
+         elif all(isinstance(spec, KpoolTailSpec) for spec in inner.values()):
+             tail_group = group
++        else:
++            draft_group = group
+     if attn_group is None or not mamba_groups:
+         return None
+-    if len(uniform_groups) + len(mamba_groups) != len(kv_cache_groups):
++    if (
++        len(uniform_groups) + len(mamba_groups) + len(hidden_groups)
++        != len(kv_cache_groups)
++    ):
+         return None
+ 
+     attn_uniform = cast(UniformTypeKVCacheSpecs, attn_group.kv_cache_spec)
+@@ -1362,6 +1411,8 @@
+         idx_page,
+         tail_names,
+         tail_page,
++        draft_group,
++        hidden_groups,
+     )
+ 
+ 
+@@ -1395,10 +1446,34 @@
+         return kv_cache_spec
+ 
+     max_page_size = max(page_sizes)
++    # GLM-5.3-Flash + a draft model: the sparse indexer k_cache page is
++    # tokens_per_state x the MLA page (same per-token bytes), and MLA layers
++    # cannot be PADDED to an arbitrary page size (sparse MLA indexes the cache
++    # in whole token rows) -- but they CAN grow their block size when the max
++    # page is an exact multiple of the MLA page. Bump the max page to a
++    # multiple of every MLA-family page so all of them take the divisible
++    # block-growth branch below instead of the padding refusal, while the
++    # draft and mamba layers simply pad to the (slightly larger) page.
++    mla_pages = {
++        spec.page_size_bytes
++        for spec in kv_cache_spec.values()
++        if isinstance(spec, MLAAttentionSpec)
++    }
++    if mla_pages:
++        mla_common = math.lcm(*mla_pages)
++        max_page_size = -(-max_page_size // mla_common) * mla_common
+     new_kv_cache_spec = {}
+     for layer_name, layer_spec in kv_cache_spec.items():
+         if layer_spec.page_size_bytes == max_page_size:
+             new_kv_cache_spec[layer_name] = layer_spec
++        elif isinstance(layer_spec, KpoolTailSpec):
++            # The kpool tail kernel pins the tail block to the kpool size
++            # (tail_kv_cache.view(...) must be [blocks, 2, index_kpool,
++            # head_dim]); growing block_size breaks that contract. Pad the
++            # physical page instead -- the tail is its own group and the pad
++            # wastes bytes only.
++            new_spec = replace(layer_spec, page_size_padded=max_page_size)
++            new_kv_cache_spec[layer_name] = new_spec
+         elif isinstance(layer_spec, MambaSpec):
+             # MambaSpec's page size is determined by its state shapes and does
+             # not scale with block_size, so pad the page instead. This is the
+@@ -1593,8 +1668,31 @@
+ ) -> int:
+     """Return the largest cache group's bytes per block."""
+     if (glm5_layout := _glm5_next_tensor_layout(kv_cache_groups)) is not None:
+-        _, _, mla_names, idx_names, mla_page, idx_page, _, _ = glm5_layout
+-        return len(mla_names) * mla_page + len(idx_names) * idx_page
++        (
++            _,
++            _,
++            mla_names,
++            idx_names,
++            mla_page,
++            idx_page,
++            _,
++            _,
++            draft_group,
++            hidden_groups,
++        ) = glm5_layout
++        bytes_per_block = len(mla_names) * mla_page + len(idx_names) * idx_page
++        if draft_group is not None:
++            d_inner = cast(
++                UniformTypeKVCacheSpecs, draft_group.kv_cache_spec
++            ).kv_cache_specs
++            bytes_per_block += sum(
++                spec.page_size_bytes for spec in d_inner.values()
++            )
++        for group in hidden_groups:
++            bytes_per_block += (
++                group.kv_cache_spec.page_size_bytes * len(group.layer_names)
++            )
++        return bytes_per_block
+ 
+     bytes_per_block = max(
+         sum(
+@@ -1701,12 +1799,30 @@
+             idx_page,
+             tail_names,
+             _,
++            draft_group,
++            hidden_groups,
+         ) = glm5_layout
+         bytes_per_block = len(mla_names) * mla_page + len(idx_names) * idx_page
++        extra_bytes_per_block = 0
++        if draft_group is not None:
++            d_inner = cast(
++                UniformTypeKVCacheSpecs, draft_group.kv_cache_spec
++            ).kv_cache_specs
++            extra_bytes_per_block += sum(
++                spec.page_size_bytes for spec in d_inner.values()
++            )
++        for group in hidden_groups:
++            extra_bytes_per_block += (
++                group.kv_cache_spec.page_size_bytes * len(group.layer_names)
++            )
+         num_blocks = may_override_num_blocks(
+-            vllm_config, available_memory // bytes_per_block
++            vllm_config,
++            available_memory // (bytes_per_block + extra_bytes_per_block),
+         )
+-        size = bytes_per_block * num_blocks
++        # One backing allocation shared by every tensor (allocate_kv_cache
++        # asserts all tensor sizes equal): the aliased GLM pool plus the
++        # draft/hidden regions appended after it.
++        size = (bytes_per_block + extra_bytes_per_block) * num_blocks
+         attn_specs = cast(
+             UniformTypeKVCacheSpecs, attn_group.kv_cache_spec
+         ).kv_cache_specs
+@@ -1745,6 +1861,37 @@
+                 ).kv_cache_specs
+                 add_tensor(tail_name, tail_specs[tail_name], offset)
+ 
++        # Draft + hidden-state layers get their own tensors appended after
++        # the aliased pool (no aliasing: independent block tables).
++        offset_cursor = bytes_per_block * num_blocks
++        if draft_group is not None:
++            d_inner = cast(
++                UniformTypeKVCacheSpecs, draft_group.kv_cache_spec
++            ).kv_cache_specs
++            d_page = next(iter(d_inner.values())).page_size_bytes
++            kv_cache_tensors.append(
++                KVCacheTensor(
++                    size=size,
++                    layers=list(draft_group.layer_names),
++                    layer_stride=d_page * num_blocks,
++                    block_stride=d_page,
++                    offset=offset_cursor,
++                )
++            )
++            offset_cursor += d_page * num_blocks * len(draft_group.layer_names)
++        for group in hidden_groups:
++            h_page = group.kv_cache_spec.page_size_bytes
++            kv_cache_tensors.append(
++                KVCacheTensor(
++                    size=size,
++                    layers=list(group.layer_names),
++                    layer_stride=h_page * num_blocks,
++                    block_stride=h_page,
++                    offset=offset_cursor,
++                )
++            )
++            offset_cursor += h_page * num_blocks * len(group.layer_names)
++
+         return KVCacheConfig(
+             num_blocks=num_blocks,
+             kv_cache_tensors=kv_cache_tensors,
+@@ -2440,6 +2587,8 @@
+             idx_page,
+             tail_names,
+             _,
++            draft_group,
++            hidden_groups,
+         ) = glm5_layout
+         uniform_spec = cast(UniformTypeKVCacheSpecs, attn_group.kv_cache_spec)
+         total_blocks = uniform_spec.max_memory_usage_pages(vllm_config)
+@@ -2452,7 +2601,30 @@
+         )
+         if tail_names:
+             total_blocks += 1
+-        return total_blocks * (len(mla_names) * mla_page + len(idx_names) * idx_page)
++        pool_bytes = total_blocks * (
++            len(mla_names) * mla_page + len(idx_names) * idx_page
++        )
++        extra_bytes = 0
++        if draft_group is not None:
++            d_inner = cast(
++                UniformTypeKVCacheSpecs, draft_group.kv_cache_spec
++            ).kv_cache_specs
++            d_spec = next(iter(d_inner.values()))
++            d_blocks = cdiv(
++                d_spec.max_memory_usage_bytes(vllm_config), d_spec.page_size_bytes
++            )
++            extra_bytes += d_blocks * d_spec.page_size_bytes * len(
++                draft_group.layer_names
++            )
++        for group in hidden_groups:
++            h_spec = group.kv_cache_spec
++            h_blocks = cdiv(
++                h_spec.max_memory_usage_bytes(vllm_config), h_spec.page_size_bytes
++            )
++            extra_bytes += (
++                h_blocks * h_spec.page_size_bytes * len(group.layer_names)
++            )
++        return pool_bytes + extra_bytes
+ 
+     bytes_per_block = _pool_bytes_per_block(kv_cache_groups)
+     total_blocks = 0

scripts/patches/glm53-dflash-04b-kv-registry-dflash.patch

@@ -0,0 +1,31 @@
+--- a/vllm/model_executor/offloader/expert_activation_registry.py
++++ b/vllm/model_executor/offloader/expert_activation_registry.py
+@@ -316,7 +316,27 @@
+ 
+ 
+ def _glm53_kv_bytes_per_token(cfg: "VllmConfig") -> int:
+-    return 13600
++    per_tok = 13600
++    # FORK-LOCAL (patches/glm53-dflash-04-kv-groups-draft): with a dflash/dspark
++    # drafter, the KV spec set gains the drafter's attention layers and the aux
++    # hidden-state cache (one bf16 hidden vector per target aux layer per
++    # token). Both are full-length per-token costs the aliased glm5 pool
++    # cannot absorb, so the reserve must include them. The drafter is a 5-layer
++    # qwen3 (8 kv heads x 128 x K+V x bf16 = 4096 B/token); the aux count is
++    # the dflash target_layer_ids count (5).
++    # NOTE: use_aux_hidden_state_outputs lives on the gpu_model_runner
++    # instance (gpu_model_runner.py:623/678), not on the spec config -- it is
++    # not set yet when the expert plan runs. The spec-config METHOD is the
++    # plan-time signal for the same fact.
++    sc = getattr(cfg, "speculative_config", None)
++    if sc is not None and getattr(sc, "method", "") in (
++        "dflash",
++        "dspark",
++        "extract_hidden_states",
++    ):
++        hidden = int(cfg.model_config.hf_config.hidden_size)
++        per_tok += 5 * hidden * 2 + 4096
++    return per_tok
+ 
+ 
+ for _arch in ("Glm5NextForConditionalGeneration", "Glm5NextForCausalLM"):

scripts/patches/gpu_worker-kv-cap.patch

@@ -0,0 +1,65 @@
+# gpu_worker-kv-cap.patch — fork-local fix for the expert-offload KV cap.
+#
+# STATUS: NOT part of the deployed frank-primary shape. It exists because the
+# fork's plan under-models this model's KV, and it is the enabler for a future
+# MTP attempt (the production GLM shape runs without MTP and boots fine on the
+# plan's own reserve).
+#
+# THE BUG: the plan models KV as 13600 B/token + a small mamba term, but
+# GLM-5.3-Flash also carries a per-sequence KDA state (34 linear-attention
+# layers) and, under MTP, the draft model's KV. gpu_worker then CAPS the KV
+# cache at that reserve, so the plan's own arithmetic can make a legal
+# --max-model-len unbootable. Measured 2026-09-30 on the pinned image:
+#   - no MTP, max_model_len 131072: reserve 1.74 GiB >= need 1.86 GiB  -> boots
+#   - MTP4,  max_model_len 131072: reserve 1.74 GiB <  need 2.07 GiB  -> refuses
+#   - MTP4,  max_model_len 262144: reserve 3.43 GiB <  need 3.62 GiB  -> refuses
+#     ("To serve at least one request with the model's max seq len ...")
+# Raising max_model_len does not fix it: the cap's slope (~13,845 B/token) only
+# overtakes the need's (~12,700 B/token + 0.52 GiB fixed) around 440k tokens,
+# where the KV reserve would eat a third of the expert arena.
+#
+# THE FIX: let an explicit operator KV request raise the cap. The compile and
+# graph-capture headroom still comes off the top, so the blast radius is one
+# flag the operator chose.
+#
+# APPLY (into a host dir that is then bind-mounted read-only over the image):
+#   IMAGE=docker.io/tcclaviger/vllm@sha256:da4029666668b90a0386c1ed23ea370c79916ad4e1a6ad1dd9db098b9fa069ea
+#   D=/home/dank/glm-patch; mkdir -p $D/vllm/v1/worker
+#   podman run --rm --entrypoint cat $IMAGE \
+#     /opt/vllm/lib/python3.14/site-packages/vllm/v1/worker/gpu_worker.py \
+#     > $D/vllm/v1/worker/gpu_worker.py
+#   ( cd $D && patch -p1 < .../scripts/patches/gpu_worker-kv-cap.patch )
+#   # then add to the podman run:
+#   #   -v $D/vllm/v1/worker/gpu_worker.py:/opt/vllm/lib/python3.14/site-packages/vllm/v1/worker/gpu_worker.py:ro
+#   #   -e CLAV_KV_RESERVE_FLOOR_GIB=<needed>   (or pass --kv-cache-memory-bytes)
+#
+# Base file sha256 (the pristine file out of the pinned image):
+#   1f3942b1a01af075317161b06077a2e658086f1da0566c9f297129ceebe37dd7
+# Patched file sha256 (verified 2026-09-30, patch applies clean and reproduces
+# these bytes exactly):
+#   80cd160bbc01270b0851f7b6fe7552b06ae2fae2428134fc29238d7f452af840
+#
+--- a/vllm/v1/worker/gpu_worker.py
++++ b/vllm/v1/worker/gpu_worker.py
+@@ -824,6 +824,21 @@
+         expert_cfg = self.vllm_config.offload_config.expert
+         if expert_cfg.enable_expert_offload and expert_cfg.planned_kv_reserve_bytes > 0:
+             cap = int(expert_cfg.planned_kv_reserve_bytes)
++            # LOCAL PATCH (GLM-5.3 KDA + MTP draft): the plan models KV as
++            # 13600 B/token + a small mamba term, but this model also carries a
++            # per-sequence KDA state and (under MTP) the draft's KV, so the
++            # reserve can sit BELOW what max_model_len legitimately needs -- and
++            # the cap then makes a legal config unbootable (measured: need
++            # 2.07 GiB vs cap 1.74 GiB at max_model_len 131072 with MTP). An
++            # explicit --kv-cache-memory-bytes, or CLAV_KV_RESERVE_FLOOR_GIB,
++            # is the operator's word: let it raise the cap. The compile/capture
++            # headroom still comes off the top.
++            _explicit = self.cache_config.kv_cache_memory_bytes
++            _floor = os.environ.get("CLAV_KV_RESERVE_FLOOR_GIB")
++            if _explicit:
++                cap = max(cap, int(_explicit))
++            if _floor:
++                cap = max(cap, int(float(_floor) * (1 << 30)))
+             if self.available_kv_cache_memory_bytes > cap:
+                 logger.info_once(
+                     "expert offload: KV capped at the plan's reserve %s GiB; %s GiB of "

scripts/serve-frank-large.sh

@@ -1,107 +1,146 @@
 #!/usr/bin/env bash
-# Serve unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL as frank-large on
-# Dankclaw's Strix Halo iGPU (gfx1151), using stock llama.cpp
-# (mainline, the unsloth b11030 prebuilt).
+# serve-frank-large.sh — frank-large endpoint: halogen (closed-source
+# C++/HIP engine, image ghcr.io/peonist-ai/halogen, digest-pinned in
+# frank-large.env) serving peonist-ai/halogen-qwen3.8-27b on Dankclaw's
+# Strix Halo iGPU, host :7770. 2026-10-05 cutover from the ROCmFPX
+# llama.cpp shape — revert the swap commit's trio to roll back (payload
+# and fork build stay on disk).
 #
-# Runtime: /home/dank/GitPackages/unsloth-llama-b11030/llama-server
-# (mainline, ROCm gfx1151 prebuilt). The build supports the qwen3_5
-# architecture and the DFlash2 drafter (llama_model_qwen35,
-# llama_model_dflash, common_speculative_impl_draft_dflash).
+# The engine is driven by `podman run` DIRECTLY (the fleet's frank-primary
+# pattern): the image entrypoint `halogen-docker` in default "all" mode
+# starts engine + OpenAI frontend in one container. The frontend listens on
+# 8731 inside (HALOGEN_API_PORT default) and is published as host 7770; the
+# engine's own token port (8730, no auth) stays unpublished on the
+# container's loopback. The image ships NO default checkpoint:
+# HALOGEN_CHECKPOINT selects the file inside the /models mount.
 #
-# Shape: one stream, full 262144 context (the model's native window), all
-# layers on the iGPU (--n-gpu-layers 999). The 27B dense model (31.46 GiB GGUF) + the
-# DFlash2 drafter (1.14 GiB) + q8_0 KV (~8.6 GiB) + compute buffers fits the
-# iGPU's GTT aperture (~114 GiB).
+# Gates mirror the deployer EXACTLY (the 2026-10-01 crash-loop lesson —
+# serve gate and deploy gate disagreeing = green deploy, red unit):
+# tokenizer directory present with every named file a FLAT REGULAR FILE —
+# the model card ships `tokenizer/` as a flat directory and the engine
+# defaults HALOGEN_TOKENIZER to /tokenizer, so a HF cache snapshot's
+# symlinks into blobs/ dangle inside the mount; GPU device nodes present.
 #
-# Spec method: DFlash2 (a block-diffusion drafter) at depth 4. The drafter
-# has a sliding window of 2048 (small draft KV, ~40 MiB).
-#
-# KV: q8_0 quantized (--cache-type-k q8_0 --cache-type-v q8_0). The unsloth b11030
-# build does not support fp8 KV (allowed: f32/f16/bf16/q8_0/q4_0/q4_1/iq4_nl/q5_0/
-# q5_1); q8_0 is the highest-precision 8-bit option and preserves the ~8 GiB footprint.
-#
-# Text-only (no mmproj) — matching the prior frank-large.
-#
-# The Qwen3.8 family template keeps prior thinking blocks across turns when
-# preserve_thinking is set; pass --jinja and the kwarg so multi-turn agents
-# do not lose the trail. The template gates reasoning_effort to {xhigh,
-# medium, low} — the harness folds its levels onto that set in models.yml
-# (reasoningEffortMap).
-#
-# Tunables come from frank-large.env unless FRANK_LARGE_ENV_FILE overrides
-# it; anything not set there falls through to the defaults below.
+# Why containers instead of the bare binary (the old frank-large shape):
+# the payload is .hgn and only halogen reads it; halogen ships as a
+# container; the root unit keeps LimitMEMLOCK=infinity for parity and to
+# outlive the login session, exactly like frank-primary's.
 set -euo pipefail
 
-SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 ENV_FILE="${FRANK_LARGE_ENV_FILE:-$SCRIPT_DIR/frank-large.env}"
-if [[ -f "$ENV_FILE" ]]; then
-    set -a
-    source "$ENV_FILE"
-    set +a
-fi
-
-# --- defaults (used when the env file does not set them) ------------------
-: "${RUNTIME_DIR:=/home/dank/GitPackages/unsloth-llama-b11030}"
-: "${MODEL_DIR:=/home/dank/models/Qwen3.8-27B-UD-Q8_K_XL}"
-: "${HOST:=0.0.0.0}"
-: "${PORT:=7770}"
-: "${CONTEXT_SIZE:=262144}"
-: "${PARALLEL_SLOTS:=1}"
-: "${SPEC_DRAFT_N_MAX:=4}"
-: "${CACHE_RAM:=8192}"
-TEMPLATE_FILE="${TEMPLATE_FILE:-$SCRIPT_DIR/qwen-sharp-v22.4.0.jinja}"
-
-SERVER_BIN="$RUNTIME_DIR/llama-server"
-MODEL_FILE="$MODEL_DIR/Qwen3.8-27B-UD-Q8_K_XL.gguf"
-DRAFTER_FILE="$MODEL_DIR/Qwen3.8-27B-DFlash2-Q4_K_M.gguf"
-
-if [[ ! -x "$SERVER_BIN" ]]; then
-    echo "runtime not found at $SERVER_BIN" >&2
+if [ ! -f "$ENV_FILE" ]; then
+    echo "frank-large: env file not found: $ENV_FILE" >&2
     exit 1
 fi
-for f in "$MODEL_FILE" "$DRAFTER_FILE"; do
-    if [[ ! -f "$f" ]]; then
-        echo "missing asset $f (run scripts/download-frank-large.sh)" >&2
+set -a
+# shellcheck disable=SC1090
+source "$ENV_FILE"
+set +a
+
+: "${FRANK_LARGE_IMAGE:?FRANK_LARGE_IMAGE must name the digest-pinned image}"
+: "${FRANK_LARGE_CONTAINER:?FRANK_LARGE_CONTAINER must name the container}"
+: "${FRANK_LARGE_PUBLISH:?FRANK_LARGE_PUBLISH must be host:container}"
+: "${FRANK_LARGE_MODEL_DIR:?FRANK_LARGE_MODEL_DIR must name the payload dir}"
+: "${FRANK_LARGE_CHECKPOINT:?FRANK_LARGE_CHECKPOINT must name the .hgn inside it}"
+: "${FRANK_LARGE_CHECKPOINT_BYTES:?FRANK_LARGE_CHECKPOINT_BYTES must pin the size}"
+: "${FRANK_LARGE_TOKENIZER_FILES:?FRANK_LARGE_TOKENIZER_FILES must list the tokenizer dir contents}"
+: "${HALOGEN_MODEL_ID:?HALOGEN_MODEL_ID must be the served alias (frank-large)}"
+: "${HALOGEN_SLOT_CTX:?HALOGEN_SLOT_CTX must be set}"
+: "${HALOGEN_KV_SLOTS:?HALOGEN_KV_SLOTS must be set (1 keeps the drafter on)}"
+: "${HALOGEN_DRAFTER:?HALOGEN_DRAFTER must be set (2 = DFlash2)}"
+
+# Capture podman's output BEFORE grepping: under `set -o pipefail` the
+# `podman images | grep -q` pattern races (grep exits on first match,
+# podman dies SIGPIPE 141, pipefail turns the 141 into "no match" — the
+# 2026-09-30 deploy refusal).
+DIGESTS="$(podman images --format '{{.RepoDigests}}')"
+if ! grep -qF "${FRANK_LARGE_IMAGE#*@}" <<<"$DIGESTS"; then
+    echo "frank-large: image $FRANK_LARGE_IMAGE not in podman storage;" >&2
+    echo "  pull it: podman pull ${FRANK_LARGE_IMAGE%@*}" >&2
+    exit 1
+fi
+
+# The checkpoint gate is size-exact: a truncated curl exits 0 on some
+# failures and a partial .hgn loads far enough to look alive, so the byte
+# count from the model card is the contract (the fleet fetch law).
+CKPT="$FRANK_LARGE_MODEL_DIR/$FRANK_LARGE_CHECKPOINT"
+if [ ! -f "$CKPT" ]; then
+    echo "frank-large: checkpoint $CKPT missing;" >&2
+    echo "  fetch it: bash scripts/download-frank-large-halogen.sh" >&2
+    exit 1
+fi
+CKPT_SIZE="$(stat -c %s "$CKPT")"
+if [ "$CKPT_SIZE" != "$FRANK_LARGE_CHECKPOINT_BYTES" ]; then
+    echo "frank-large: checkpoint is $CKPT_SIZE bytes, expected $FRANK_LARGE_CHECKPOINT_BYTES (partial download?)" >&2
+    exit 1
+fi
+
+# The payload ships the tokenizer as a FLAT DIRECTORY (model card:
+# "tokenizer/ flat tokenizer directory"). A HuggingFace cache snapshot
+# symlinks each file into a sibling blobs/, and those symlinks dangle
+# inside the container mount — so the directory must be real and every
+# named file must be a regular file, not a link.
+TOK_DIR="$FRANK_LARGE_MODEL_DIR/tokenizer"
+if [ -L "$TOK_DIR" ] || [ ! -d "$TOK_DIR" ]; then
+    echo "frank-large: tokenizer dir $TOK_DIR missing or a symlink (HF cache snapshot?);" >&2
+    echo "  fetch it: bash scripts/download-frank-large-halogen.sh" >&2
+    exit 1
+fi
+for TF in $FRANK_LARGE_TOKENIZER_FILES; do
+    if [ ! -f "$TOK_DIR/$TF" ] || [ -L "$TOK_DIR/$TF" ]; then
+        echo "frank-large: tokenizer file $TOK_DIR/$TF missing or a symlink;" >&2
+        echo "  fetch it: bash scripts/download-frank-large-halogen.sh" >&2
         exit 1
     fi
 done
 
-# --- env (the unsloth build's ROCm setup) --------------------------------
-export HIP_VISIBLE_DEVICES=0
-export HSA_OVERRIDE_GFX_VERSION=11.5.1
-export LD_LIBRARY_PATH="/opt/rocm/lib:/opt/rocm/llvm/lib:${LD_LIBRARY_PATH:-}"
+# GPU access is mandatory but the node paths are variables so the seam can
+# drive the gate on a box without an AMD GPU.
+DEV_KFD="${FRANK_LARGE_DEV_KFD:-/dev/kfd}"
+DEV_DRI="${FRANK_LARGE_DEV_DRI:-/dev/dri}"
+if [ ! -e "$DEV_KFD" ] || [ ! -d "$DEV_DRI" ]; then
+    echo "frank-large: $DEV_KFD or $DEV_DRI missing — no gfx1151 to serve on" >&2
+    exit 1
+fi
+for CTR in ${FRANK_LARGE_LEGACY_CONTAINERS:-} "$FRANK_LARGE_CONTAINER"; do
+    podman rm -f "$CTR" >/dev/null 2>&1 || true
+done
 
-# --- the serve command ---------------------------------------------------
-ARGS=(
-    --host "$HOST"
-    --port "$PORT"
-    --model "$MODEL_FILE"
-    --alias frank-large
-    --ctx-size "$CONTEXT_SIZE"
-    --parallel "$PARALLEL_SLOTS"
-    --n-gpu-layers 999
-    --jinja
-    --metrics
-    --chat-template-file "$TEMPLATE_FILE"
-    --chat-template-kwargs '{"preserve_thinking":true}'
-    --cache-type-k q8_0
-    --cache-type-v q8_0
-    --cache-ram "$CACHE_RAM"
-    # DFlash2 spec decode (a block-diffusion drafter)
-    --spec-type draft-dflash
-    --model-draft "$DRAFTER_FILE"
-    --spec-draft-n-max "$SPEC_DRAFT_N_MAX"
-    --spec-draft-n-min 0
-    --spec-draft-device ROCm0
-    # Greedy sampling (the DFlash2 drafter needs greedy verification)
-    --temp 0
-    --top-p 1.0
-    --top-k 0
-    --min-p 0.0
-    # Reasoning (thoughts land in reasoning_content)
-    --reasoning auto
-    --reasoning-format deepseek
-    --reasoning-budget -1
-)
+LOG="${FRANK_LARGE_LOG:-/home/dank/logs/frank_large.log}"
+mkdir -p "$(dirname "$LOG")"
+# for the udev-granted supplementary group, seccomp unconfined for the ROCm
+# ioctl surface, and unbounded memlock (the unit's LimitMEMLOCK=infinity is
+# inherited, so runc never has to RAISE it). NO --ulimit nofile: Dankclaw's
+# hard limit is 524288, rootless runc cannot raise a hard limit and dies
+# with "error setting rlimit type 7: operation not permitted" (measured
+# 2026-10-05 cutover), and 524288 fds is already ~100x what the engine
+# wants; the model card and both sibling serve scripts pass no nofile
+# either. No comment lines inside the backslash
+# chain — a mid-chain # terminates the continuation (the 2026-10-05
+# exit-127 crash loop on the sibling serve script).
+podman run -d --name "$FRANK_LARGE_CONTAINER" \
+  --publish "$FRANK_LARGE_PUBLISH" \
+  --ipc=host \
+  --ulimit memlock=-1:-1 \
+  --device "$DEV_KFD" --device "$DEV_DRI" \
+  --group-add keep-groups \
+  --security-opt seccomp=unconfined \
+  -e HALOGEN_MODEL_ID="$HALOGEN_MODEL_ID" \
+  -e HALOGEN_CHECKPOINT="/models/$FRANK_LARGE_CHECKPOINT" \
+  -e HALOGEN_SLOT_CTX="$HALOGEN_SLOT_CTX" \
+  -e HALOGEN_KV_SLOTS="${HALOGEN_KV_SLOTS:-1}" \
+  -e HALOGEN_DRAFTER="$HALOGEN_DRAFTER" \
+  -e HALOGEN_MAX_TOKENS_CAP="${HALOGEN_MAX_TOKENS_CAP:-65536}" \
+  -e HALOGEN_QUEUE_TIMEOUT="${HALOGEN_QUEUE_TIMEOUT:-7200}" \
+  ${HALOGEN_CACHE_MB:+-e HALOGEN_CACHE_MB="$HALOGEN_CACHE_MB"} \
+  -v "$FRANK_LARGE_MODEL_DIR:/models:ro" \
+  -v "$FRANK_LARGE_MODEL_DIR/tokenizer:/tokenizer:ro" \
+  "$FRANK_LARGE_IMAGE" \
+  > "$LOG" 2>&1
 
-exec "$SERVER_BIN" "${ARGS[@]}"
+# This script's lifetime IS the container's: podman wait returns the
+# container's exit code, so a crash is a nonzero ExecStart and the unit's
+# Restart=always is real. systemd stop goes through the unit's ExecStop
+# (podman wait alone cannot stop the child).
+exec podman wait "$FRANK_LARGE_CONTAINER"

scripts/serve-frank-primary.sh

@@ -1,30 +1,31 @@
 #!/usr/bin/env bash
-# serve-frank-primary.sh — frank-primary endpoint (R9V radiance-vLLM fork,
-# unsloth Qwen3.8-Flash-Next Q8_0 + MTP4 drafter, 2x R9700, dankserver
-# :7770; 2026-09-27 cutover from the uncensored IQ4_XS payload, whose state
-# dir stays on disk as the rollback target).
-#
-# The engine is the upstream R9V launcher (scripts/launch.sh in the fork
-# checkout), NOT a hand-frozen podman argv: launch.sh owns the resolved
-# docker/podman args (binds, overlays, device passthrough, publish lines)
-# and the fork's qualification was measured through it. This script's job:
-#
-#   1. layer the on-host R9V setup state (the 104-key tuned shape the
-#      profile setup qualified) under this repo's frank-primary.env
-#      overrides (bind/port/alias/container);
-#   2. gate on the pinned runtime image (by its OCI manifest digest, the
-#      podman RepoDigest), the exact-size PLE payload, and the GGUF shards;
-#   3. exec the launcher, then wait on the container: launch.sh --detach
-#      exits as soon as the container is created, so without the wait the
-#      unit would be "green" while the engine crashes — the wait is what
-#      keeps systemd's Restart=on-failure self-healing real.
+# serve-frank-primary.sh — frank-primary endpoint (FreeToken PR #587 ROCm
+# build, orcarouter Qwen3.8-Flash-Next-Uncensored-FP8, ONE R9700 gfx1201,
+# dankserver :7770; 2026-10-05 cutover from the tcclaviger vLLM fork — git
+# tag frank-primary-vllm-shapeG-20261003 is the pre-cutover state).
 #
+# The engine is driven by `podman run` DIRECTLY and the run line below IS
+# the engine argv: the image carries FreeToken installed at pinned PR head
+# 77a7201 (scripts/docker/frank-primary-freetoken.Dockerfile builds it from
+# the PR's own docs/install_amd.md stack). Unlike the vLLM fork there is no
+# per-boot planner: experts live in pinned host RAM with an LRU GPU cache,
+# KV is an explicit token pool, and the boot ledger in the log names
+# Shape: 1 GPU (first-boot conservative — the second R9700 sits idle;
+# --tensor-parallel-size EXISTS in FreeToken, so a 2-card TP shape is future
+# tuning once this single-card shape has a decode baseline), KV pool 300,000
+# tokens (24 KiB/token f16 across the 12 full-attention layers = 7.2 GiB;
+# clears the harness worst case of 177,356 tokens and admits one NATIVE
+# 262,144-token conversation), MoE LRU at 10% of all experts (~12 GiB),
+# 2 decode slots. First-boot conservative; `ft ctl cache --kv/--moe` grows
+# pools live, and the env comment block carries the resize law. Flag
+# surface verified against docs/cli.md of the pinned head AND against
+# `ft serve --help` inside the built image; every value here is env-driven
+# (frank-primary.env).
+
 # The root unit (scripts/frank-primary.service, User=dank,
-# LimitMEMLOCK=infinity) grants the pin budget for the tiered-expert UVA
-# host copy and the PLE pinned residency (54.4 GB on the q8 payload). A
-# plain shell session also works (proven 2026-09-27: podman rootless
-# inherits usable limits here), but the unit is the fleet-parity
-# boot-persistent shape.
+# LimitMEMLOCK=infinity) grants the pin budget: ~120 GiB of expert weights
+# plus the ~47.7 GiB PLE table pin in host RAM. A plain shell session also
+# works, but the unit is the fleet-parity boot-persistent shape.
 set -euo pipefail
 
 SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
@@ -39,104 +40,93 @@ set -a
 source "$ENV_FILE"
 set +a
 
-: "${R9V_HOME:?R9V_HOME must name the fork checkout}"
-: "${R9V_STATE_FILE:?R9V_STATE_FILE must name the setup state}"
-: "${R9V_HOST_BIND:?R9V_HOST_BIND must be set}"
-: "${R9V_HOST_PORT:?R9V_HOST_PORT must be set}"
-: "${R9V_SERVED_MODEL_NAME:?R9V_SERVED_MODEL_NAME must be set}"
-: "${R9V_CONTAINER_NAME:?R9V_CONTAINER_NAME must be set}"
+: "${FRANK_PRIMARY_IMAGE:?FRANK_PRIMARY_IMAGE must name the FreeToken image}"
+: "${FRANK_PRIMARY_MODEL_DIR:?FRANK_PRIMARY_MODEL_DIR must name the FP8 payload}"
+: "${FRANK_PRIMARY_MODEL_SHARDS:?FRANK_PRIMARY_MODEL_SHARDS must give the shard count}"
+: "${FRANK_PRIMARY_CONTAINER:?FRANK_PRIMARY_CONTAINER must name the container}"
+: "${FRANK_PRIMARY_SERVED_NAME:?FRANK_PRIMARY_SERVED_NAME must name the alias}"
+: "${FRANK_PRIMARY_PUBLISH:?FRANK_PRIMARY_PUBLISH must be host:container}"
+: "${FRANK_PRIMARY_FT_GPU:?FRANK_PRIMARY_FT_GPU must name the GPU index}"
+: "${FRANK_PRIMARY_FT_NUM_TOKENS:?FRANK_PRIMARY_FT_NUM_TOKENS must size the KV pool}"
+: "${FRANK_PRIMARY_FT_MAX_RUNNING:?FRANK_PRIMARY_FT_MAX_RUNNING must be set}"
+: "${FRANK_PRIMARY_FT_MAX_OUTPUT_TOKENS:?FRANK_PRIMARY_FT_MAX_OUTPUT_TOKENS must be set}"
+: "${FRANK_PRIMARY_FT_MOE_RATE:?FRANK_PRIMARY_FT_MOE_RATE must be set}"
 
-LAUNCHER="$R9V_HOME/scripts/launch.sh"
-if [ ! -x "$LAUNCHER" ]; then
-    echo "serve-frank-primary: R9V launcher missing: $LAUNCHER" >&2
+# The image gate is EXISTENCE IN STORAGE, not a registry digest: the tag is
+# built ON THE HOST (podman build from
+# scripts/docker/frank-primary-freetoken.Dockerfile) and local builds carry
+# no RepoDigests entry — the 362-line vLLM-era RepoDigests grep refused
+# exactly this shape. Pinning lives in the Dockerfile instead: the FreeToken
+# source is checked out at a hardcoded PR-head SHA and the base image is
+# digest-pinned there. `podman image exists` accepts both a local tag and a
+# name@digest ref, and it is one command — no pipefail/grep race (the
+# 2026-09-30 lesson) for either shape.
+if ! podman image exists "$FRANK_PRIMARY_IMAGE"; then
+    echo "serve-frank-primary: image absent from rootless storage: $FRANK_PRIMARY_IMAGE" >&2
+    echo "  build it: podman build -t $FRANK_PRIMARY_IMAGE -f scripts/docker/frank-primary-freetoken.Dockerfile scripts/docker/" >&2
     exit 1
 fi
 
-# The docker->podman shim is REQUIRED: the launcher and its doctor speak
-# docker (OCI-manifest-digest image IDs, LogConfig fields) that podman's
-# native CLI answers differently. The shim carries those laws.
-SHIM_DIR="$R9V_HOME/.podman-shim"
-if [ ! -x "$SHIM_DIR/docker" ]; then
-    echo "serve-frank-primary: docker->podman shim missing: $SHIM_DIR/docker" >&2
+if [ ! -f "$FRANK_PRIMARY_MODEL_DIR/config.json" ]; then
+    echo "serve-frank-primary: payload config missing: $FRANK_PRIMARY_MODEL_DIR/config.json" >&2
     exit 1
 fi
-export PATH="$SHIM_DIR:$PATH"
-
-# --- layer 1: the qualified setup state (production overrides never come
-# from here; keys already defined in the env file are skipped) -------------
-STATE_FILE="$R9V_STATE_FILE"
-if [ ! -f "$STATE_FILE" ]; then
-    echo "serve-frank-primary: R9V setup state missing: $STATE_FILE" >&2
-    echo "  run the profile setup first (tools/setup_profile.py setup --state-dir ...)" >&2
+SHARDS="$(find "$FRANK_PRIMARY_MODEL_DIR" -maxdepth 1 -name 'model-*-of-*.safetensors' | wc -l)"
+if [ "$SHARDS" != "$FRANK_PRIMARY_MODEL_SHARDS" ]; then
+    echo "serve-frank-primary: payload shard count wrong: $FRANK_PRIMARY_MODEL_DIR" >&2
+    echo "  found $SHARDS safetensors, want $FRANK_PRIMARY_MODEL_SHARDS" >&2
+    echo "  re-fetch: scripts/download-frank-primary.sh" >&2
     exit 1
 fi
-STATE_EXPORTS="$(python3 - "$STATE_FILE" <<'PY'
-import json, shlex, sys, os
-state = json.load(open(sys.argv[1]))
-config = state.get("config") or {}
-if state.get("ready") is not True:
-    sys.exit("R9V setup state is not ready; rerun setup before serving")
-env_keys = {k for k in os.environ if k.startswith("R9V_")}
-for key in sorted(config):
-    if key.startswith("R9V_") and key not in env_keys:
-        print(f"export {key}={shlex.quote(config[key])}")
-PY
-)"
-eval "$STATE_EXPORTS"
-
-: "${R9V_IMAGE:?state must carry R9V_IMAGE}"
-: "${R9V_MODEL_DIR:?state must carry R9V_MODEL_DIR}"
-: "${R9V_PLE_PATH:?state must carry R9V_PLE_PATH}"
-
-# --- gates: never serve a half-finished payload or a missing image --------
-# Runtime image by pinned OCI manifest digest (podman keys images by config
-# digest but preserves the manifest digest in RepoDigests — the shim's ID law).
-if ! podman images --format '{{.RepoDigests}}' | grep -qF "${R9V_IMAGE#sha256:}"; then
-    echo "serve-frank-primary: pinned R9V image absent: $R9V_IMAGE" >&2
-    echo "  load the pinned image bundle into podman rootless storage" >&2
+if [ ! -f "$FRANK_PRIMARY_MODEL_DIR/model.safetensors.index.json" ]; then
+    echo "serve-frank-primary: weight-map index missing: $FRANK_PRIMARY_MODEL_DIR/model.safetensors.index.json" >&2
     exit 1
 fi
 
-# PLE payload at its pinned byte count (stat -c %s, never wc -c), plus the
-# six target GGUF shards (mradermacher-style split parts) and the mmproj
-# the launcher mounts.
-PLE_BYTES="${R9V_PLE_EXPECTED_BYTES:-}"
-if [ -n "$PLE_BYTES" ]; then
-    ACTUAL="$(stat -c %s "$R9V_PLE_PATH" 2>/dev/null || echo 0)"
-    if [ "$ACTUAL" != "$PLE_BYTES" ]; then
-        echo "serve-frank-primary: PLE payload wrong size: $R9V_PLE_PATH" >&2
-        echo "  expected $PLE_BYTES, found $ACTUAL" >&2
-        exit 1
-    fi
-else
-    if [ ! -f "$R9V_PLE_PATH" ]; then
-        echo "serve-frank-primary: PLE payload missing: $R9V_PLE_PATH" >&2
-        exit 1
-    fi
-fi
-for rel in "$R9V_TARGET_REL" "${R9V_TARGET_SHARD2_REL:-}" "${R9V_TARGET_SHARD3_REL:-}" \
-    "${R9V_TARGET_SHARD4_REL:-}" "${R9V_TARGET_SHARD5_REL:-}" "${R9V_TARGET_SHARD6_REL:-}" \
-    "$R9V_MMPROJ_REL"; do
-    if [ -z "$rel" ]; then continue; fi
-    if [ ! -f "$R9V_MODEL_DIR/$rel" ]; then
-        echo "serve-frank-primary: payload file missing: $R9V_MODEL_DIR/$rel" >&2
-        exit 1
-    fi
-done
-
 if [ ! -e /dev/kfd ] || [ ! -d /dev/dri ]; then
     echo "serve-frank-primary: GPU device nodes absent (/dev/kfd, /dev/dri)" >&2
     exit 1
 fi
 
 # A crashed/killed unit can leave the container name registered (systemd
-# SIGKILLs conmon before cleanup runs); the launcher refuses an existing
-# name forever after. Clear any orphan of OUR name first — idempotent.
-docker rm -f "$R9V_CONTAINER_NAME" >/dev/null 2>&1 || true
+# SIGKILLs conmon before cleanup runs). Clear OUR name plus the legacy
+# identities (the vLLM-era frank-primary-qwen, the retired R9V container and
+# the ad hoc experiment containers all hold this GPU); all idempotent.
+for CTR in $FRANK_PRIMARY_LEGACY_CONTAINERS "$FRANK_PRIMARY_CONTAINER"; do
+    podman rm -f "$CTR" >/dev/null 2>&1 || true
+done
 
-# launch.sh detaches after preflight; from here on, this script's lifetime
-# IS the container's: wait returns the container's exit code, so a crash is
-# a nonzero ExecStart and Restart=on-failure fires. (systemd stop goes
-# through ExecStop in the unit — docker wait alone cannot stop the child.)
-"$LAUNCHER"
-exec docker wait "$R9V_CONTAINER_NAME"
+LOG=/home/dank/logs/frank_primary.log
+mkdir -p "$(dirname "$LOG")"
+
+# --device /dev/dri is the WHOLE DIRECTORY (individual renderD nodes enumerate
+# without the card nodes and HIP refuses contexts). --ipc=host for the
+# engine's shared memory; memlock unbounded for the ~120 GiB pinned expert
+# rows + the 47.7 GiB PLE pin (the unit's LimitMEMLOCK=infinity is the other
+# half of that law).
+# HIP_VISIBLE_DEVICES pins visibility to ONE card so in-container index 0 is
+# the right R9700; --gpu 0 selects it inside the visible set.
+podman run -d --name "$FRANK_PRIMARY_CONTAINER" \
+  --publish "$FRANK_PRIMARY_PUBLISH" \
+  --ipc=host --ulimit memlock=-1:-1 \
+  --cap-add SYS_PTRACE \
+  --device /dev/kfd --device /dev/dri --device /dev/mem \
+  -e HIP_VISIBLE_DEVICES="$FRANK_PRIMARY_FT_GPU" \
+  -v "$FRANK_PRIMARY_MODEL_DIR":/models:ro \
+  "$FRANK_PRIMARY_IMAGE" \
+  ft serve --model /models \
+    --served-model-name "$FRANK_PRIMARY_SERVED_NAME" \
+    --host 0.0.0.0 --port 7770 \
+    --gpu "$FRANK_PRIMARY_FT_GPU" \
+    --num-tokens "$FRANK_PRIMARY_FT_NUM_TOKENS" \
+    --max-running-requests "$FRANK_PRIMARY_FT_MAX_RUNNING" \
+    --max-output-tokens "$FRANK_PRIMARY_FT_MAX_OUTPUT_TOKENS" \
+    --moe-cache-rate "$FRANK_PRIMARY_FT_MOE_RATE" \
+    ${FRANK_PRIMARY_FT_EXTRA_ARGS:+$FRANK_PRIMARY_FT_EXTRA_ARGS} \
+  > "$LOG" 2>&1
+
+# From here on, this script's lifetime IS the container's: wait returns the
+# container's exit code, so a crash is a nonzero ExecStart and the unit's
+# Restart=always is real. (systemd stop goes through ExecStop in the unit —
+# podman wait alone cannot stop the container child.)
+exec podman wait "$FRANK_PRIMARY_CONTAINER"

scripts/serve-frank-small.sh

@@ -1,39 +1,25 @@
 #!/usr/bin/env bash
-# Serve HauhauCS/Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced (Q4_K_M) as
-# frank-small on Dankclaw's RTX 3090 via mainline llama.cpp.
+# Serve ornith-ai/Ornith-1.5-9B-GGUF (Q8_0) as frank-small on
+# Dankclaw's RTX 3090 via llama.cpp (master build, CUDA sm_86).
 #
-# Topology: the whole model lives in VRAM (-ngl 999) -- 6.87 GiB of Q4_K_M
-# weights plus the BF16 vision projector and q8_0 KV inside the card's
-# 24 GiB. The host CPU only tokenizes and samples here; this box's CPU
-# belongs to the halogen engine that serves frank-large.
+# 2026-10-01 night (Frank): swapped from SGLang + INT8-W8A16-AutoRound
+# safetensors to the llama.cpp GGUF path, because llama.cpp is the only
+# runtime on this host that moves K and V independently — the asymmetric
+# f16-K + q8_0-V contract that endpoint was built for. 2026-10-03 (Frank):
+# the pair is now f16/f16 — two wide slots freed the VRAM, so the values are
+# unquantized too. The per-tensor flags stay explicit either way.
 #
-# MTP: this model's Multi-Token Prediction head ships as a SEPARATE file
-# (mtp-gemma-4-12B-it.gguf, arch gemma4-assistant, 4 nextn layers). It is
-# passed via --model-draft with --spec-type draft-mtp: the draft context
-# loads the assistant model. The same --spec-type value chains the
-# draftless ngram-mod matcher, which attaches unconditionally; two
-# context-backed drafters cannot share one process, so this chain is the
-# only shape that carries both.
+# Architecture: qwen3_5 hybrid — 32 layers, 8 full-attention + 24 Gated
+# DeltaNet, native 262144 window. KV lives only in the 8 attention layers;
+# the DeltaNet state is per-slot fixed size. Vision comes from the repo's
+# separate projector (the Q6_K GGUF carries no `v.*` tensors at all), so
+# --mmproj is what gives this endpoint eyes; MMPROJ_FILE="" is the documented
+# text-only opt-out.
 #
-# Context: CTX_SIZE is set explicitly to PARALLEL x 262144 so every slot
-# gets the model's FULL window; --fit (the CTX_SIZE-unset branch below)
-# maximises the TOTAL pool instead, which would divide a smaller window
-# across slots. The 4-slot layout is measured to fit the card (see the
-# SIZING note in frank-small.env); more slots would fail the KV allocation.
+# Sampling: llama.cpp takes it per request (the fleet harness sends its
+# recipe from models.yml); greedy is the harness default.
 #
-# Vision + audio: --mmproj loads the BF16 projector, which also carries
-# an audio encoder — input_audio content parts transcribe verbatim
-# (verified 2026-09-17). The fleet's image path lives here, which is why
-# halogen no longer sets HALOGEN_VISION_TOWER.
-#
-# Sampling: greedy by default (the fleet contract). Clients that send their own
-# temperature get their own values, and the harness overrides per model from
-# models.yml. --jinja is pinned even though this build defaults it on: the
-# model's own template supplies reasoning_effort, preserve_thinking and the
-# tool-call grammar, and a future build that flips that default must not
-# silently take tools away.
-#
-# Tunables come from frank-small.env unless FRANK_SMALL_ENV_FILE overrides it.
+# Tunables come from frank-small.env unless FRANK_SMALL_ENV_FILE overrides.
 set -euo pipefail
 
 SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
@@ -46,95 +32,68 @@ if [[ -f "$ENV_FILE" ]]; then
 fi
 
 # --- defaults (used when the env file does not set them) -------------------
-: "${LLAMA_BIN:=/home/dank/GitPackages/llama.cpp-mainline/build-cuda/bin/llama-server}"
-: "${MODEL_DIR:=/home/dank/models/Qwen3.8-27B-GSQ-RCO-IQ3_S-mtp}"
-: "${MODEL_FILE:=Qwen3.8-27B-GSQ-RCO-IQ3_S-mtp.gguf}"
-: "${MMPROJ_PATH:=$MODEL_DIR/mmproj-Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced-BF16.gguf}"
-: "${DRAFT_FILE:=}"
+: "${LLAMA_BIN:=/home/dank/GitPackages/llama-master/build-cuda/bin/llama-server}"
+: "${MODEL_FILE:=/home/dank/models/Ornith-Q8/Ornith-1.5-9B-Q8_0.gguf}"
 : "${ALIAS:=frank-small}"
 : "${BIND_ADDR:=0.0.0.0}"
 : "${LISTEN_PORT:=7771}"
-: "${CTX_SIZE:=}"
-: "${PARALLEL:=1}"
+# ctx-size is the POOL llama.cpp divides across slots: 401408 / 2 gives
+# 200,704 per slot — two independent conversations, each one wide enough to
+# hold a whole frank-primary-sized prompt in a single shot (the native 262,144
+# no longer fits twice on a 24 GiB card once the vision tower is loaded;
+# frank-small.env carries the three measurements that set this number).
+: "${CTX_SIZE:=401408}"
+: "${PARALLEL:=2}"
+# Compute-buffer geometry. Before 2026-10-03 this endpoint never passed either
+# flag, so it ran on llama.cpp's own defaults (2048/512). The pair is 4096/2048
+# since 2026-10-05: on this CUDA card the prefill curve climbs all the way to
+# ubatch 2048 (unlike the R9700 layer-split sibling, where 512 is the peak),
+# and the logical batch is VRAM- and throughput-neutral. frank-small.env
+# carries the sweep tables and the VRAM risk note.
+: "${BATCH_SIZE:=4096}"
+: "${UBATCH_SIZE:=2048}"
 : "${CACHE_TYPE_K:=f16}"
 : "${CACHE_TYPE_V:=f16}"
-: "${CACHE_RAM:=8192}"
-: "${BATCH_SIZE:=2048}"
-: "${UBATCH_SIZE:=512}"
-# Decode-shape knobs: the MTP drafter's per-round proposal length (build
-# default 3; the Gemma4 sidecar carries 4 nextn layers) and the engine's
-# busy-wait level (build default 50).
-: "${SPEC_DRAFT_N_MAX:=3}"
-: "${POLL_LEVEL:=50}"
-: "${NUM_THREADS:=8}"
-: "${THREAD_BATCH:=8}"
-: "${NGL:=999}"
-: "${FLASH_ATTN:=on}"
-: "${SPEC_TYPE:=draft-mtp}"
-: "${TEMPERATURE:=0.0}"
-: "${REASONING_FORMAT:=deepseek}"
-: "${REASONING_BUDGET:=-1}"
+: "${NGL:=99}"
+# Vision projector. No colon on purpose: an explicitly empty value means
+# "text-only, on purpose" and must NOT be replaced by the default path.
+: "${MMPROJ_FILE=/home/dank/models/Ornith-Q8/mmproj-Ornith-1.5-9B-BF16.gguf}"
 
-# Fail loudly rather than half-loading a model into VRAM: a missing file here
-# is a download that has not finished, and the unit's restart policy would
-# otherwise turn that into a crash loop that reads like a runtime fault.
-for path in "$LLAMA_BIN" "$MODEL_DIR/$MODEL_FILE" "$MMPROJ_PATH"; do
+# Fail loudly rather than half-loading: a missing file is a download that
+# has not finished, and Restart=always would turn it into a crash loop that
+# reads like a runtime fault.
+for path in "$LLAMA_BIN" "$MODEL_FILE"; do
     [[ -e "$path" ]] || { echo "frank-small: missing $path" >&2; exit 1; }
 done
-if [[ -n "${DRAFT_FILE:-}" ]]; then
-    [[ -e "$MODEL_DIR/$DRAFT_FILE" ]] || { echo "frank-small: missing $MODEL_DIR/$DRAFT_FILE" >&2; exit 1; }
+EXTRA_ARGS=()
+if [[ -n "$MMPROJ_FILE" ]]; then
+    [[ -e "$MMPROJ_FILE" ]] || { echo "frank-small: missing $MMPROJ_FILE" >&2; exit 1; }
+    EXTRA_ARGS+=(--mmproj "$MMPROJ_FILE")
 fi
 
 # The UVM device node is created lazily by nvidia-modprobe on the first CUDA
 # open after boot; make sure it exists before the server needs it.
 nvidia-modprobe -u -c=0 2>/dev/null || true
 
-MMPROJ_ARGS=()
-if [[ -n "${MMPROJ_PATH:-}" && -f "$MMPROJ_PATH" ]]; then
-    MMPROJ_ARGS=(--mmproj "$MMPROJ_PATH")
-fi
-
-SPEC_ARGS=()
-if [[ -n "${SPEC_TYPE:-}" && "$SPEC_TYPE" != "none" ]]; then
-    SPEC_ARGS=(--spec-type "$SPEC_TYPE" --spec-draft-n-max "$SPEC_DRAFT_N_MAX")
-fi
-
-DRAFT_ARGS=()
-if [[ -n "${DRAFT_FILE:-}" ]]; then
-    DRAFT_ARGS=(--model-draft "$MODEL_DIR/$DRAFT_FILE")
-fi
-
-# CTX_SIZE set -> pinned context. Unset -> --fit sizes the KV pool against
-# whatever the weights, projector, draft context and compute buffers leave
-# on the card; the engine prints the chosen value at startup.
-if [[ -n "${CTX_SIZE:-}" ]]; then
-    CTX_ARGS=(--ctx-size "$CTX_SIZE")
-else
-    CTX_ARGS=(--fit on)
-fi
-
-exec "$LLAMA_BIN" \
-    --model "$MODEL_DIR/$MODEL_FILE" \
-    "${MMPROJ_ARGS[@]}" \
-    "${DRAFT_ARGS[@]}" \
-    --alias "$ALIAS" \
-    --host "$BIND_ADDR" \
-    --port "$LISTEN_PORT" \
-    --gpu-layers "$NGL" \
-    --flash-attn "$FLASH_ATTN" \
-    "${CTX_ARGS[@]}" \
-    --parallel "$PARALLEL" \
-    --batch-size "$BATCH_SIZE" \
-    --ubatch-size "$UBATCH_SIZE" \
-    --threads "$NUM_THREADS" \
-    --threads-batch "$THREAD_BATCH" \
-    --poll "$POLL_LEVEL" \
-    --cache-type-k "$CACHE_TYPE_K" \
-    --cache-type-v "$CACHE_TYPE_V" \
-    --cache-ram "$CACHE_RAM" \
-    "${SPEC_ARGS[@]}" \
-    --jinja \
-    --metrics \
-    --temp "$TEMPERATURE" \
-    --reasoning-format "$REASONING_FORMAT" \
-    --reasoning-budget "$REASONING_BUDGET"
+# Flash attention stays pinned on: it is the fast path for f16 KV, and it is
+# the only path that could serve quantized V at all — so if CACHE_TYPE_V is
+# ever set back to q8_0 this flag must not be lost, or the build falls back
+# silently and the KV contract in frank-small.env becomes a lie.
+ARGS=(
+    --model "$MODEL_FILE"
+    --alias "$ALIAS"
+    --host "$BIND_ADDR"
+    --port "$LISTEN_PORT"
+    --ctx-size "$CTX_SIZE"
+    --parallel "$PARALLEL"
+    --batch-size "$BATCH_SIZE"
+    --ubatch-size "$UBATCH_SIZE"
+    --n-gpu-layers "$NGL"
+    --flash-attn on
+    --cache-type-k "$CACHE_TYPE_K"
+    --cache-type-v "$CACHE_TYPE_V"
+    --jinja
+    --reasoning auto
+    --reasoning-format deepseek
+)
+exec "$LLAMA_BIN" "${ARGS[@]}" ${EXTRA_ARGS[@]+"${EXTRA_ARGS[@]}"}

serverAppConfigs-docs/Agent/Agent Docs and Edits.md

@@ -0,0 +1,111 @@
+---
+type: reference
+title: Agent Docs and Edits
+description: Two self-inflicted damage classes — structurally broken markdown after parallel agent edits, and botched hashline PUT/CUT repairs — with the audit checks and recovery order for each.
+tags: [agents, docs, markdown, hashline, edit-recovery]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Agent Docs and Edits
+
+## Doc-structure audit (after parallel agent edits)
+
+Use after one or more subagents have edited markdown docs (OKF/Obsidian
+vault with `---` frontmatter), before committing. Link checkers and
+completeness seams validate *references*, not *structure* — agents reliably
+produce structurally broken markdown that still links cleanly.
+
+Damage catalogue:
+
+1. Banner/prose prepended **above** the opening `---` → the frontmatter is
+   no longer frontmatter (title/tags/description become body text).
+2. A rewrite **appended** instead of merged → duplicate headings and a stray
+   second `---` mid-file.
+3. Literal model artifacts in the body: `<|channel>}<tool_call|>...`,
+   `+- `, `- - `.
+4. Wrapped bullets re-emitted with a `- ` prefix on every physical line.
+5. Line-range replacement that dropped a sentence tail → dangling fragment,
+   often mid-parenthesis.
+
+Checks (fast, run all):
+
+```bash
+head -3 <file>                          # must be `---` if the page has frontmatter
+grep -c '^---$' <file>                  # exactly 2 (opener + closer)
+grep -n '^## ' <file> | sort -t: -k2 | uniq -f1 -d   # duplicate sections
+grep -rnE '^\+- |^- - |<\|' <files>     # artifacts
+python3 scripts/check-vault-links.py <vault-dir>
+./tests/run.sh                          # docs completeness seam etc.
+```
+
+Repair order (matters):
+
+1. If the file was destroyed rather than edited: stash the agent version to
+   `.AGENTTRASH/<YYYYMMDD-topic>/`, then `git checkout HEAD -- <file>` and
+   re-apply only the intended changes by hand.
+2. If it merely duplicated: **salvage unique content first** — cut the
+   one-of-a-kind section, paste it into the copy you keep, THEN delete the
+   duplicate ranges. Never delete first.
+3. Re-add missing sentence tails (insert before the orphaned fragment),
+   keeping list continuation lines indented.
+4. Re-run all checks above.
+
+Gotchas: agents can write AFTER they appear finished (job settles, file
+changes) — re-verify structure after the last background job reports, not
+before. Editing an agent's already-damaged file with fresh line numbers is
+safe, but its snapshot tag changes on every edit — re-read before each
+subsequent edit. `.AGENTTRASH/` is gitignored: trash is local-only, so
+nothing moved there reaches the repo — that is intended (temporary hold, not
+archive).
+
+## Hashline edit recovery
+
+Failure mode: multi-op edits resolve every `PUT N.=M:` / `CUT N.=M:` index
+against the **original** file numbering, and the post-apply diff view
+**elides lines** (it shows an anchor window, not the whole file). Two traps
+follow:
+
+1. A second edit call built from a diff-view snapshot uses line numbers from
+   a *partial* view. Lines outside the window are invisible but still
+   counted.
+2. If the first apply was itself wrong, the file shifted. The diff view
+   tells you the damage, but not the true offsets.
+
+Result: the repair pastes fresh text **beside** stale text instead of over
+it.
+
+Procedure:
+
+1. **Never carry line numbers across calls.** Re-read the target range for a
+   fresh tag before every subsequent edit. A read is cheap; a duplicated
+   paragraph costs a commit and a revert.
+2. **One op per call when the previous apply was imperfect.** Fixed offsets
+   are only trustworthy when every preceding apply landed exactly as
+   specified.
+3. **Re-read after every apply**, and verify the whole affected region — the
+   diff view is a sample, not proof. Check the *boundaries*: the line before
+   the edit and the line after it. Damage shows up as a repeated sentence
+   straddling the seam.
+4. **Repair by CUT, not by re-PUT.** Once you know the exact stale lines
+   from a fresh read, delete them. Do not try to PUT a corrected superset
+   over a range whose extent you are guessing at.
+5. Verify with a content assertion, not a diff review: `grep -c` for the
+   duplicated sentence, plus a syntax/compile check on any touched script.
+
+Same rule for any imperative `edit` tool with an old/new match: the matched
+text must be the **entire** text being replaced. Matching a bare fragment
+inside a paragraph splices the replacement into the middle of the sentence
+and silently corrupts the file. Re-read the file afterwards to confirm.
+
+### Anchored in-page links
+
+Renaming a Markdown heading breaks `[[#Old Heading]]` anchors. When a
+section is retitled, grep for the old heading text and fix every anchor in
+the same pass — doc link-checkers frequently do not validate `#fragment`
+targets, so a green suite will not catch a dangling one.
+
+## Related
+
+- [[OMP Harness]] — the session context these repairs protect
+- [[Command Center Dotfiles]] — the repo these edits commit to
+- [[Testing Policy]] — the seams that catch broken vault structure

serverAppConfigs-docs/Agent/Command Center Dotfiles.md

@@ -0,0 +1,87 @@
+---
+type: reference
+title: Command Center Dotfiles
+description: The ~/.omp repo is itself the deployment — stow-equivalent packages, full-replica rsync transport, the pre-overwrite audit for non-blank hosts, deploy.sh mechanics, and the fish/sshpass transport gotchas.
+tags: [dotfiles, omp, deploy, rsync, blankpad]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Command Center Dotfiles
+
+Repo `~/.omp` is itself the deployment: packages are home-relative trees
+under `dotfiles/<pkg>/`; `scripts/deploy.sh` symlinks live files into the
+repo checkout. `git pull` + `deploy.sh` is the whole propagation story.
+
+## Transport decision (agreed with Frank, executed 2026-09-09)
+
+- Ship via **full-replica rsync of `/home/dank/.omp/` including `.git`** to
+  `<user>@<host>:/home/<user>/.omp/`, NOT clone-from-Gitea — `main` can be
+  ahead of `origin/main`; push is a user-only trigger, so clone would ship
+  stale state.
+- Deploy runs on-target as the target user: `bash
+  /home/<user>/.omp/scripts/deploy.sh` (resolves the repo via `git -C
+  $(dirname $(realpath $0)) rev-parse --show-toplevel`, so `.git` must be
+  present at the target path — full-replica transport guarantees this).
+
+## Pre-overwrite audit (MANDATORY before first sync on a non-blank host)
+
+Diff every file that `deploy.sh --check` reports as DRIFT against the repo
+copy BEFORE letting quarantine-and-replace run. Host-specific divergence
+belongs in an override package, not in quarantine:
+
+1. Pull divergent live files from target
+   (`ssh ... 'cat <path>' > local-staged-copy`).
+2. Secret-audit staged copies (grep token/secret/password/key shapes)
+   before anything enters git.
+3. Compose `dotfiles/zz-<hostname>/` on the source machine from those live
+   copies; commit to repo first.
+4. Ship the package dir via rsync, then run deploy sync — override deploys
+   LAST and win over base packages.
+
+## deploy.sh mechanics (verified on blankpad)
+
+- `--check` = pure report (ADOPT/LINK/DRIFT lines), writes nothing. Run it
+  first, always.
+- Default mode = sync: adopt identicals silently, link new,
+  quarantine-and-replace divergents into `.quarantine/<host>/<ts>/`
+  (gitignored; never rm).
+- **Override-target bug (fixed cc2c652):** `zz-<host>` packages nest
+  base-package dirs internally (`zz-blankpad/niri/.config/...`); target
+  derivation must strip the nested package segment, else overrides deploy to
+  phantom `$HOME/niri/.config/...` while real targets keep pointing at
+  canonical copies (visible symptom: check shows 0 actions but
+  `readlink ~/.zshrc` points at base `zsh/.zshrc`).
+- systemd units auto-enable only when ExecStart binary resolves on that
+  host; dconf packages load via `dconf load /`, never symlinks.
+- `core.hooksPath=.githooks` is armed by deploy.sh per machine — set it once
+  post-transfer even before first sync.
+
+## ssh/sshpass transport gotchas (all bit live)
+
+- Resolve the target via mDNS FQDN (`avahi-resolve -4 -n <host>.local`); a
+  bare hostname may not resolve. Drop-not-refuse on port 22 = sshd down or
+  firewall default-drop, not auth failure — needs a local/console fix on
+  target.
+- `SSHPASS='...' sshpass -e ssh ...` inline per invocation; an env passed
+  via the tool's env parameter did not propagate to the sshpass child
+  process.
+- The remote login shell may be **fish**: never send bash syntax
+  (`for/do/done`, `$?`) bare through `ssh host 'cmd'`; wrap as `bash -c
+  "..."` or ship a script. See [[SSH and Remote Access]].
+- File shipping: `ssh 'cat > /remote/path' < localfile` (stdin redirect). A
+  here-doc attached to the ssh command is consumed by `cat` — the intended
+  script body becomes the file content, silently. Verify shipped bytes with
+  an md5sum compare on both ends; do NOT compare `/dev/stdin` after `cat`
+  drained it (false DIVERGED).
+
+## Verification of a completed deploy
+
+Re-run `deploy.sh --check` → expect exactly zero actions/quarantined/skipped
+= idempotent steady state. Spot-check `readlink` on override files points
+into `dotfiles/zz-<host>/`, not base packages.
+
+## Related
+
+- [[SSH and Remote Access]] — the transport mechanics this deploy rides
+- [[Noctalia and Niri]] — the desktop packages this repo ships
+- [[Deploy State Triage]] — the serverAppConfig deploy flow (different repo)

serverAppConfigs-docs/Agent/OMP Harness.md

@@ -0,0 +1,262 @@
+---
+type: reference
+title: OMP Harness
+description: The omp agent harness as configured on this fleet — compaction chain arithmetic, per-model sampler routing via compat.extraBody, and the static-preamble/skill-library audits that keep session context lean.
+tags: [omp, harness, compaction, models-yml, context, samplers]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# OMP Harness
+
+The harness reaches the fleet through `~/.omp/agent/models.yml` (model
+registry; single source of truth — there is no copy on the servers) and
+`~/.omp/agent/config.yml` (roles, compaction, global samplers). Nothing in
+this file is repo-managed; edits are harness-side only.
+
+## Compaction chain arithmetic (verify before ANY window change)
+
+When a fleet endpoint's context grows past its harness-declared window, OMP
+compacts by sending the conversation to a designated target model. Before
+changing any `contextWindow`/`maxTokens`, verify the chain arithmetically —
+a target whose window is smaller than the source's trigger point silently
+breaks compaction.
+
+Where the contract lives:
+
+- `~/.omp/agent/config.yml` → `compaction.remoteEndpoint` (target URL),
+  `methodOrder`, `thresholdPercent` (70), idle settings
+- `~/.omp/agent/models.yml` → per-model `contextWindow`, `maxTokens`,
+  `compactionModel` references
+- Implementation: omp plugin `dist/core/compaction/compaction.js`
+
+The arithmetic (the coupled contract every endpoint's maxTokens comment
+encodes):
+
+1. **Source trigger** = `thresholdPercent × source contextWindow`
+   (0.70 × 200,704 = 140,492 for frank-primary on the current shape)
+2. **Output budget** = capped at the target model's `maxTokens`
+   (32,768 for frank-small; 36,864 frank-primary; 57,344 frank-large — the
+   largest, because halogen's `HALOGEN_MAX_TOKENS_CAP` must stay ≥ it)
+3. **REQUIREMENT**: prompt (≤ trigger) + budget ≤ the target's per-slot
+   window. Worked pairs from the live registry: 140,492 + 36,864 = 177,356 ≤
+   200,704 (frank-primary); 140,490 + 57,344 = 197,834 ≤ 200,700
+   (frank-large); 140,492 + 32,768 = 173,260 ≤ 200,704 (frank-small).
+
+Procedure (recompute from the LIVE files every time — windows rotate with
+every swap; the numbers above are the 2026-10-05 state):
+
+1. Read the `config.yml` compaction block + `models.yml` entries for source
+   and target. Every endpoint's `contextWindow` comment carries its own
+   arithmetic — read them.
+2. Read the target checkpoint's `max_position_embeddings` (native window) —
+   the served window may be a concurrency contract, not the native limit
+   (`--parallel N` DIVIDES the pool; the per-slot number is the contract).
+3. If native ≥ trigger + budget, reslice the pinned pool:
+   `MAX_RUNNING × CONTEXT_LENGTH ≤ MAX_TOTAL_TOKENS` (the pin exists because
+   the runtime's autofit leaves too little for deferred loads + prefill
+   workspace; check `nvidia-smi` headroom — expect ~0.8 GB workspace margin).
+4. Changes: host env (CONTEXT_LENGTH-style env tunables) via git →
+   `./tests/run.sh` → deploy; harness `models.yml` directly (no repo copy).
+5. Verify: `/health` 200, boot log's pool line, then a long-prompt probe at
+   ≥ trigger height (the compaction-fit guarantee), then concurrent requests.
+
+Known reconciliation trap: comments in models.yml have described compaction
+as running "on the dankserver endpoint" while `remoteEndpoint` pointed at
+dankclaw — the comment/config disagreed. The comments now name the real
+target (`compaction.remoteEndpoint` = dankclaw :7771). **The config is
+truth; fix comments in the same change.** `maxTokens` never changes for a
+reslice — it caps output; `contextWindow` caps input. Boot-shape OOM
+laddering is a different problem ([[Llama Cpp Serving Sizing]]).
+
+## Per-model sampler config (`compat.extraBody`)
+
+`models.yml` has **no per-model sampler keys**. Sampling is global
+(`config.yml`: `temperature`, `topP`, `topK`, `minP`, `presencePenalty`,
+`repetitionPenalty`; default `-1` = **not sent**). The per-model hook is
+`compat.extraBody`:
+
+**Why it works:** `applyOpenAIExtraBody` runs **last** in the
+openai-completions builder — after `applyChatCompletionsCompatPolicy` — via
+`Object.assign(params, extraBody)`. So extraBody overrides global settings
+and compat-policy output.
+
+```yaml
+compat:
+  reasoningEffortMap:
+    max: xhigh
+  extraBody:
+    temperature: 1.0
+    top_p: 0.95
+    top_k: 20
+    min_p: 0.0
+    presence_penalty: 0.0
+    repetition_penalty: 1.0
+    preserve_thinking: true
+```
+
+Use **wire names**, not OMP camelCase. Two traps:
+
+1. **`preserve_thinking` is not user-settable as a compat key.**
+   `qwenPreserveThinking` is auto-derived in `buildOpenAICompat` from the
+   *auto-detected* `thinkingFormat` + local baseUrl, **before**
+   `applyCompatOverrides`. Setting `compat.thinkingFormat: qwen` in
+   models.yml does **not** enable it. Custom-named models (`frank-large`)
+   match no Qwen id pattern → use `extraBody: {preserve_thinking: true}`.
+2. **Never put `chat_template_kwargs` in `extraBody`** where a compat policy
+   also sets it — `Object.assign` is shallow and replaces the whole object,
+   silently dropping the `reasoning_effort` kwarg the compat policy set.
+   Top-level `preserve_thinking` is honoured by halogen. (frank-primary's
+   FreeToken entry currently carries a `chat_template_kwargs` extraBody with
+   a comment asking to verify-and-prune — the exact trap to check at that
+   endpoint's next swap.)
+
+**Gotcha: silent greedy decode.** halogen has **no server-side sampler
+defaults** — `temperature` absent or `0` = greedy. If OMP sends nothing, the
+endpoint decodes greedy while looking configured. This is why frank-large's
+entry carries the recipe on the wire and a comment saying "do not drop it".
+
+**Reasoning-effort folding.** Qwen3.8's template accepts only
+`{xhigh (default), medium, low}` for `reasoning_effort` and 400s on anything
+else ("Unexpected reasoning effort high"). Every endpoint's
+`reasoningEffortMap` folds all harness levels onto that set; a role whose
+level is unmapped fails on its FIRST request, and the failure names Jinja,
+not the model.
+
+### Verify the exact wire body (echo-capture)
+
+Don't trust the YAML — capture what OMP actually emits:
+
+1. Echo server on `127.0.0.1:18099` that dumps the JSON body to a file and
+   returns a canned `chat.completion` (include `reasoning_content` so
+   thinking paths exercise).
+2. Copy the **real** models.yml into a throwaway profile with only the
+   baseUrl swapped:
+   ```bash
+   sed 's|http://<host>:7770/v1|http://127.0.0.1:18099/v1|' \
+     ~/.omp/agent/models.yml > ~/.omp/profiles/wiretest/agent/models.yml
+   ```
+3. Run and inspect:
+   ```bash
+   omp --profile wiretest --model "<provider>/<model>" -p "hi" \
+     --no-tools --no-session --no-extensions --no-skills --no-rules
+   jq 'del(.messages, .tools)' last-body.json
+   ```
+   Confirm every intended field is present with the right value.
+
+**Verify a field is *honoured*, not just accepted** — unknown fields often
+return 200 anyway. Prove behavior with a **prompt-token delta**: replay an
+assistant turn carrying a long `reasoning_content`, then compare
+`usage.prompt_tokens` with the flag true vs false. A large delta means the
+field changed template rendering; zero delta means it was ignored.
+
+Cleanup: trash the probe profile and echo server (`profiles/.AGENTTRASH`),
+never `rm`.
+
+## Static-preamble audit ("the context feels bloated")
+
+What loads statically every session (measure each, don't guess):
+
+1. Repo rules file — `wc -c AGENTS.md` (baseline 2026-10-05: 72,179 B, ~18k
+   tokens — historically the largest block).
+2. Skill index — only names+descriptions load; bodies are on-demand. Sum
+   description bytes across all SKILL.md frontmatter:
+   ```bash
+   python3 - <<'EOF'
+   import glob, re, os
+   for root in [os.path.expanduser("~/.omp/agent/managed-skills"),
+                os.path.expanduser("~/.omp/agent/skills"), ".omp/skills"]:
+       n=tot=0
+       for sk in glob.glob(root+"/*/SKILL.md"):
+           fm=re.search(r'^---.*?---', open(sk, encoding="utf-8", errors="replace").read(), re.S)
+           d=re.search(r'description:\s*(.*(?:\n(?:\s{2,}.*)*)?)', fm.group(0) if fm else "")
+           n+=1; tot+=len(d.group(1)) if d else 0
+       print(root, n, tot)
+   EOF
+   ```
+   Baseline 2026-10-05: 171 managed skills / ~38 KB descriptions + 9 user
+   skills.
+3. Memory lessons — `find ~/.omp/agent/memories -name 'MEMORY.md' -o -name
+   'memory_summary.md' | xargs wc -c`. The injected block ≈ the largest
+   MEMORY.md.
+4. Harness/system prompt + tool schemas + CLI persona: fixed cost, ~20 KB;
+   not prunable.
+
+Known rot patterns:
+
+- Near-duplicate auto-captures of the same lesson (three "authentik ORM"
+  entries, the freetoken-build pair, the three job-runner skills) — dedupe
+  via `manage_skill delete`/`update`.
+- Corrupted captures containing stray CJK characters (`一团ly`, `等方法`)
+  mid-sentence — a known defect of the auto-capture path; delete the corrupt
+  copy, keep the clean one.
+- Stale host/service names in consolidated memory (pre-rename
+  `frank-default-large`) — the repo's AGENTS.md topology table is ground
+  truth.
+
+Fix strategy: AGENTS.md prose (measured runbook narratives) → move to
+`serverAppConfigs-docs/LLM/` runbooks, keep laws + tables in AGENTS.md (the
+repo's own docs-tree rule mandates this). Managed skills live in
+`~/.omp/agent/managed-skills` (safe to edit); user-authored skills in
+`~/.omp/agent/skills` are NEVER touched by tooling. Memory store edits:
+stage removed captures to trash first (`~/.AGENTTRASH/YYYYMMDD-topic`).
+
+Baseline to beat (2026-10-05): ~35–40k tokens of preamble — AGENTS.md ~18k,
+skill index ~10k, lessons ~3.5k, harness ~6k. Proposed reductions: −12k
+(AGENTS.md slimming), −3k (skill dedupe).
+
+## Managed-skill library hygiene
+
+Use when auditing, pruning, or consolidating `~/.omp/agent/managed-skills`
+(the auto-captured skill store; the full index is injected into EVERY
+session's context, so dead or duplicate entries are a permanent tax).
+
+The static-context budget (measured 2026-10-05):
+
+| Block | Size | Notes |
+|---|---|---|
+| `AGENTS.md` (repo) | 72,179 B ≈ 18k tok | biggest single item |
+| Skill index (~173 skills) | ~40 KB ≈ 10k tok | names+descriptions only |
+| Learned-lessons block | ~14 KB | can contain corrupted/duplicate captures |
+| Harness system prompt + tool schemas | ~23 KB | fixed |
+
+Audit procedure:
+
+1. Build an inventory: per skill, name, mtime, body word count, description,
+   first 12 headings (one pass, out to a file; read in pages).
+2. Group into: retired-runtime (host/service gone; its runbook already lives
+   in `serverAppConfigs-docs/LLM/`), verbatim-duplicate pairs/triples (same
+   lesson captured twice — often the SAME DAY), same-lesson families
+   (N skills = one procedure), keep.
+3. Read full bodies of every merge candidate before proposing a merge;
+   description alone proves overlap only for exact dupes.
+4. Present A/B/C/D verdicts for user approval. NEVER execute before the
+   discussion concludes — the user gates execution explicitly.
+5. On approval: stage everything to trash first
+   (`~/.AGENTTRASH/YYYYMMDD-<topic>/`), never rm. After merging: recreate
+   the surviving skill via `manage_skill` update with body = superset,
+   delete absorbed ones.
+
+Known rot classes (this library, 2026-10-05):
+
+- Same-day duplicate captures (freetoken-build-dankclaw vs
+  freetoken-dankclaw-build; the three job-runner skills).
+- CUDA-skill vs its self-declared HIP twin (pcie-link-capability-probe vs
+  pcie-link-probe-hip-rdna4) → merge.
+- Era ghosts: skills naming retired runtimes (freetoken, lucebox, r9v/
+  radiance, old halogen-flash-server image, ik_llama, GLM-5.3 experiment).
+- Skills contradicting user standing orders (live-bench-dashboard vs the
+  2026-09-29 delete-dashboards order) — flag for user, don't assume.
+- Memory store: corrupted captures with CJK noise from broken ssh commands;
+  stale host/service renames.
+
+Sizing verdicts that worked: bench family → 2 skills (harness vs
+discipline), not one mega-skill; keep profiler/kernel-time attribution
+standalone; fold go/no-go arithmetic skills into their probe skill
+(heterogeneous-accelerator-split-triage → PCIe probe).
+
+## Related
+
+- [[Agent Docs and Edits]] — editing vault docs and hashline targets safely
+- [[Command Center Dotfiles]] — the `~/.omp` repo deploy this config rides
+- [[Llama Cpp Serving Sizing]] — the engine-side sizing the chain must fit
+- [[LLM Bench Windows and Measurement]] — measuring the endpoints this config routes

serverAppConfigs-docs/Agent/index.md

@@ -0,0 +1,23 @@
+---
+type: index
+description: Pages about the agent harness itself — omp config, vault editing discipline, and the dotfiles repo that deploys the agent's own home.
+tags: [agent, index]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Agent
+
+Pages about the agent's own tooling and context — how omp is configured, how
+docs get edited safely, and how the agent's home is deployed.
+
+- [[OMP Harness]] — compaction chain arithmetic, per-model samplers via
+  `compat.extraBody`, preamble and skill-library audits
+- [[Agent Docs and Edits]] — doc-structure audits after parallel edits;
+  hashline edit recovery
+- [[Command Center Dotfiles]] — the `~/.omp` repo-as-deployment, rsync
+  transport, override packages
+
+## Related
+
+- [[Testing Policy]] — the seams these pages' checks ride
+- [LLM](../LLM/index.md) — the fleet these agents operate

serverAppConfigs-docs/Architecture/Deploy State Triage.md

@@ -0,0 +1,151 @@
+---
+type: reference
+title: Deploy State Triage
+description: Reading and reconciling the deploy-state markers — stale marker vs live container, pending-service lockout, interrupted/raced deploys — without restarting healthy heavy endpoints.
+tags: [deployment, deploy-state, systemd, podman, triage]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Deploy State Triage
+
+The deploy-state store is `~/.local/state/serverAppConfigs/deploy-state/<svc>`
+on each target (rootless, user `dank`, format exactly two lines
+`revision=<sha>` / `status=deployed|pending`, written atomically via tmp file
++ `mv`). It is written ONLY by a `deploy.sh` dispatch.
+
+## 1. Stale marker vs live container
+
+**Why it happens:** an out-of-band relaunch — reboot, `systemctl restart`, or
+a restart after the checkout ff-pulled to a newer revision — starts the
+container from the CURRENT `scripts/<svc>.env` but does NOT advance the
+marker. So `marker != HEAD` means "not dispatched via deploy.sh", NOT
+"behavior pending".
+
+**Why it matters:** a plain `deploy.sh` run re-selects and restarts a service
+whose marker lags. On llama.cpp/R9V endpoints that is a ~16–19 min reload of
+a 100+ GiB payload — and frank-primary (dankserver:7770) is also the
+harness's own model backend, so a blind deploy is a self-inflicted outage for
+possibly zero behavioral change.
+
+**Procedure (read-only first):**
+
+1. Committed state: `git log --oneline main..<branch>` empty ⇒ merged; the
+   operational delta is usually one `scripts/<svc>.env`.
+2. Probe the target (ssh as `dank`, passwordless sudo present):
+   - `cat ~/.local/state/serverAppConfigs/deploy-state/<svc>`
+   - `git -C ~/serverAppConfigs rev-parse HEAD`
+   - `systemctl is-active/is-enabled <svc>`;
+     `curl -s -o /dev/null -w '%{http_code}' http://127.0.0.1:<port>/health`
+   - the on-host env file already carries the new keys (it landed via the
+     ff-pull).
+3. **Decisive — read the LIVE container, not the marker.** Use `podman` as
+   `dank` (rootless store; bare `docker` is NOT the shim → "No such object";
+   `sudo podman` inspects an empty store):
+   - `podman inspect <ctr> --format '{{.State.StartedAt}}'` — booted AFTER
+     the pull ⇒ it read the new env
+   - `podman inspect <ctr> --format '{{.Args}}' | tr ' ' '\n' | grep -i <flag>`
+   - engine env: `podman exec <ctr> sh -c 'for f in /proc/[0-9]*/environ; do
+     tr "\0" "\n" <$f 2>/dev/null; done | grep -i <VAR>'`
+   - confirm the launcher maps each tunable to its flag
+     (`grep scripts/launch.sh`), e.g. `R9V_ENABLE_PREFIX_CACHING=1 →
+     --enable-prefix-caching`.
+
+**Decision:** live argv+env == target config AND `/health` 200 ⇒ behavior is
+live — **do NOT restart**; reconcile the marker (below). Only if the live
+container lacks the flag/env (or predates the pull) is a real
+`deploy.sh <target> --service <svc>` warranted.
+
+**Reconcile (no restart)** — mirror the deployer's atomic idiom:
+
+```bash
+S=$HOME/.local/state/serverAppConfigs/deploy-state
+H=$(git -C ~/serverAppConfigs rev-parse HEAD)
+t=$(mktemp "$S/.<svc>.XXXXXX"); printf 'revision=%s\nstatus=%s\n' "$H" deployed > "$t"; mv -f -- "$t" "$S/<svc>"
+```
+
+Then prove the gate: `./tests/run.sh` + `tests/e2e/run.sh` (needs
+`tests/e2e/.env.local`; expect the documented `11-admin` red if the
+FrankBook-Pro Mac mini 192.168.0.112 is ASLEEP — confirm with `ping -c2` +
+a `/dev/tcp` probe: 100 % loss + timeout = host off, not a regression).
+Log to `serverAppConfigs-docs/log.md`; commit + push (docs-only ⇒ deploy
+no-op).
+
+**Anti-patterns:** trusting a commit message ("X deployed") or the marker
+alone — both are bookkeeping; the container's argv/env is ground truth.
+Reading container state with `sudo podman` / bare `docker`. Restarting a
+healthy heavy endpoint to satisfy a stale marker.
+
+## 2. Pending lockout
+
+Symptom: `deploy.sh <target>` fails or silently stops with
+`service <name> is not deployed on <target>` and the `[3/3]` e2e gate never
+appears.
+
+1. **Get the real exit code first:** `./scripts/deploy.sh X 2>&1 | tail`
+   reports TAIL's 0 — use `set -o pipefail` or redirect to a file, then
+   `echo $?`.
+2. Inspect the state store (no sudo):
+   `ssh <target> 'cat ~/.local/state/serverAppConfigs/deploy-state/<name>'`.
+   A `status=pending` entry for a service absent from `service_is_allowed`
+   in `scripts/deploy-remote.sh` = retired-service lockout.
+3. Fixed behavior (since 02a7876, 2026-09-17): `add_pending_services` prunes
+   not-allowed pending entries to
+   `~/.AGENTTRASH/<YYYYMMDD>-retired-deploy-state/<service>` and the run
+   continues. Verify the WARN line names the service and the state file is
+   gone.
+4. On a host without the fix: move the stale entry to
+   `~/.AGENTTRASH/<date>-retired-deploy-state/` manually (never `rm`), re-run.
+5. Verify: deploy reaches `[3/3] Running Playwright e2e gate` and completes;
+   `deploy-state/` no longer holds the retired name.
+
+Related contracts: the superseded seam
+(`deployment_scoping.bats` "remote runner recognizes a pending service at
+the same revision") now asserts prune-and-continue (exit 0), not exit 2;
+`deployment_flow.bats` extracts `add_pending_services` via sed and
+HOME-overrides the trash path. **When retiring a service going forward:**
+delete or reconcile its deploy-state entry as part of the retirement commit —
+the prune is a safety net, not the intended cleanup path.
+
+## 3. Interrupted or raced deploys
+
+The deploy is TWO processes: `deploy.sh` (dev box) and `deploy-remote.sh`
+(runs on the target over ssh). **Killing the dev-side job does NOT kill the
+host-side one** — two `deploy-remote.sh` processes can end up managing the
+same unit, one deploying a revision you already superseded. Check:
+`pgrep -af "deploy-remote"` on the target, read the revision argument on
+each, kill stale ones by explicit PID.
+
+A killed deploy leaves behind:
+
+- the unit **stopped/failed** (`Restart=always` does NOT fire after an
+  explicit stop — start it yourself);
+- the state marker **pending** (pre-deploy or in-flight revision);
+- the e2e gate un-run (the deployer runs it last).
+
+Reconcile the marker ONLY after the service itself is verified healthy — a
+marker that runs ahead of reality is worse than one that lags (the planner
+uses it to decide what is pending). Then close the loop:
+
+1. Verify against MEASURED, config-specific numbers (KV capacity line, arena
+   slots, argv flags, `NRestarts`), not just health 200.
+2. Run the gate the dead deploy skipped (`tests/e2e/run.sh` from a checkout
+   that has the gitignored `tests/e2e/.env.local` — a fresh `git worktree`
+   does NOT have it; copy it in first).
+3. If the killed revision was restoring a known-good shape, confirm the
+   numbers match the last measured-good boot of that shape — equality is the
+   proof, not health alone.
+
+Boot failures must be read from the journal, not `podman logs`: serve scripts
+`podman rm -f` the container on every start (previous attempt lost), and
+`podman logs` on a container still *Stopping* returns the OLD configuration's
+numbers — which looks exactly like your change having no effect. Key on
+container identity (`podman ps -a --format '{{.CreatedAt}}'`) and read the
+cause from `journalctl --since "N min ago" | grep -a "<container-name>\["`.
+See [[vLLM Boot Forensics and Config Interrogation]] for the full capture
+procedure.
+
+## Related
+
+- [[Deployment Flow]] — the deploy pipeline itself
+- [[Quadlet Lifecycle]] — unit ownership and drift diagnosis
+- [[vLLM Boot Forensics and Config Interrogation]] — journal capture

serverAppConfigs-docs/Architecture/E2E Suite.md

@@ -20,11 +20,12 @@ user sees it; the `bats` seams assert config shape, not behavior.
 
 | Path | Role |
 |------|------|
-| `playwright.config.ts` | four projects (below), real-browser UA, trace on failure |
+| `playwright.config.ts` | five projects (below), real-browser UA, trace on failure |
 | `specs/00-auth.setup.ts` | setup project: two real authentik logins, saves storage states |
 | `specs/10-homepage.spec.ts` | user panel set + open links derived from `services.json` (registry-derived — no hardcoded service lists) |
 | `specs/11-admin.spec.ts` | `admin` project: admin grid + `/admin/` RBAC (403 for standard user, members-only 200) |
 | `specs/12-crawl.spec.ts` | every user open link reaches its real app through the guard chain |
+| `specs/19-frankbook.spec.ts` | `frankbook` project: walks the six admin links that terminate on FrankBook-Pro (192.168.0.112) — conditional, see below |
 | `specs/13-sso.spec.ts` | direct guarded-vhost visits resolve silently with ONLY the homepage session (single-login enforcement) |
 | `specs/14-guard.spec.ts` | `guest` project, no storage state: unauthenticated traffic lands on the IdP login |
 | `lib/login.ts` | `akLogin` helper (two-stage flow, shadow-DOM locators) |
@@ -38,7 +39,16 @@ user sees it; the `bats` seams assert config shape, not behavior.
   `.auth/qa-e2e{,-admin}.json` storage states.
 - **standard** — 10/12/13 with the user state.
 - **admin** — 11 with the admin state.
-- **guest** — 14 with an empty state.
+- **guest** — 14/15 with an empty state.
+- **frankbook** — 19 with the admin state. **Excluded from every default run**: the
+  *arr/qBittorrent links terminate on FrankBook-Pro, which sleeps when idle, and a
+  sleeping Mac mini must never fail an unrelated deploy. The project joins only
+  when `E2E_INCLUDE_FRANKBOOK=1` (set by `scripts/deploy.sh` when a frankbook
+  trigger path changed: `tests/e2e/specs/19-frankbook.spec.ts`, `frankbook-pro/**`,
+  `shared/homepage/services.json`, or the `dankclaw/Caddyfile` serving the arr
+  vhosts) or when an explicit `--project` argument names it. An `11-admin` edit
+  that touches the arr links must therefore be paired with a `19-frankbook` run
+  against an awake host before it is trusted live.
 
 Credentials live in `tests/e2e/.env.local` (git-ignored; placeholders in
 `.env.local.example`). The scaffold bats seam fails the commit if creds or
@@ -47,8 +57,9 @@ auth states are ever tracked.
 ## Run
 
 ```bash
-cd tests/e2e && npx playwright test            # all projects
-npx playwright test --project=standard 12-crawl.spec.ts   # one slice
+tests/e2e/run.sh                          # default set — frankbook project excluded
+E2E_INCLUDE_FRANKBOOK=1 tests/e2e/run.sh  # default set + the frankbook walk
+cd tests/e2e && npx playwright test       # ALL projects, frankbook included
 ```
 
 Requires LAN reachability of `www.franksdank.cloud` (Pi-hole DNS or public).

serverAppConfigs-docs/Architecture/Quadlet Lifecycle.md

@@ -0,0 +1,209 @@
+---
+type: reference
+title: Quadlet Lifecycle
+description: The full Quadlet discipline on this fleet — ownership auditing, compose-to-Quadlet cutover, drift/crash-loop diagnosis, the podman 5.x v1 generator quirks, log retrieval, and compose bring-up with ground truth.
+tags: [podman, quadlet, systemd, boot-persistence, compose, drift]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Quadlet Lifecycle
+
+Fleet rule (AGENTS.md "Boot Persistence & Self-Healing"): every deployed,
+working container MUST have a self-healing, boot-persistent Quadlet unit —
+`.container` file in the repo (`<host>/<svc>/quadlet/`), installed by the
+deployer, `[Install] WantedBy=default.target`, `Restart=always`, linger on.
+A container that runs only from compose dies silently at reboot. This page
+is the operating manual for that lifecycle.
+
+## 1. Ownership audit — find the containers that will not survive a reboot
+
+```bash
+for c in $(podman ps --format '{{.Names}}'); do
+  printf '%-26s %s\n' "$c" "$(podman inspect $c --format '{{index .Config.Labels "PODMAN_SYSTEMD_UNIT"}}')"
+done
+```
+
+- **label present** → the unit owns it; persistence = `[Install]` + `loginctl
+  show-user <user> -p Linger` = yes.
+- **label absent, compose service exists** → the classic gap: dies at reboot.
+  Check the publish address while there — compose forms often bind the
+  wildcard (an unauthenticated UI/API on every interface).
+- **label present AND a compose service with the same name** → two owners; a
+  stray `podman compose up` can create a duplicate that fights the unit for
+  the name and the port. Remove the compose service, leave an explanatory
+  comment (repo convention).
+
+## 2. Cutover: compose service → Quadlet unit
+
+Checklist (precedents to copy:
+`dankclaw/{caddy,gitea,searxng,open-webui,seerr}/quadlet/`,
+`dankclaw/observability/quadlet/`):
+
+1. Unit file at `<host>/<svc>/quadlet/<name>.container`, mirroring compose
+   line for line: `Image=` (`localhost/` prefix for locally built images +
+   `Pull=never`), `ContainerName=` (Caddy and scrape configs address
+   containers by name), `Network=`, `User=0:0` when the image's own user
+   cannot write a rootless-owned bind mount, `Volume=` (`:ro` preserved),
+   `Exec=` for compose `command:`, `PublishPort=`, health keys,
+   `[Service] Restart=always` + `RestartSec=10`, `[Install]
+   WantedBy=default.target`. **Never `systemctl --user enable` the generated
+   unit** — `[Install]` + linger IS the persistence (`is-enabled` = `generated`
+   is healthy; explicit enable fails with "transient or generated").
+2. Exposure: publish on `127.0.0.1` unless something off-host genuinely needs
+   it (Caddy reaches containers by name over the shared network; the wildcard
+   publish is a guard-chain bypass).
+3. Handler in `scripts/deploy-remote.sh`: `install_quadlet <dir> <name>` →
+   `systemctl --user restart <name>.service` → `wait_for_health`; retire any
+   legacy hand-written unit into `~/.AGENTTRASH/` (a leftover
+   `~/.config/systemd/user/<svc>.service` that shells out to
+   `podman compose up` collides BY NAME with the generated unit);
+   `podman rm -f <container>` so the unit's container can take the name.
+4. Planner: map `<host>/<svc>/**` → service in `scripts/deploy-plan.py`
+   AND add a `service_is_allowed` entry; keep `SERVICE_ORDER` dependencies
+   ordered. Add a `test-changed` case above the generic
+   `*/quadlet/*.container` fallback.
+5. Retire the compose entry in the SAME change — two owners is how a stale
+   duplicate comes back.
+6. Seam test `tests/<svc>_quadlet.bats`: boot persistence, restart policy, no
+   wildcard publish, container name matches Caddy/scrape references, health
+   command, deployer installs+restarts+waits, never `rm`, never `enable`.
+   Migrate any older assertion that read the compose file for devices/flags.
+7. Deploy `--service <name>`; prove: `is-active`/`is-enabled` =
+   active/generated, container `healthy`, endpoint answers,
+   `list-dependencies default.target` shows the unit, off-host probes on
+   previously published ports now refuse, and the container no longer carries
+   `io.podman.compose.project`.
+
+**Traps that cost real time:**
+
+- `HealthCmd` = CMD-SHELL (podman runs `/bin/sh -c`); a distroless image has
+  no shell → the check fails forever. `--health-cmd` won't take exec form
+  either. Put `HEALTHCHECK … CMD ["binary","arg"]` in the image and build
+  with `--format docker` — the default OCI format has no HEALTHCHECK field
+  and drops it with only a warning.
+- Never point a unit's `EnvironmentFile` at the repo-root `.env` — it is a
+  gluetun env carrying `HOME=/root`; systemd sets HOME for the service and
+  the unit dies with `stat /root/.config: permission denied`. Use a
+  per-service env under `/home/dank/Containers/<svc>/env`, seeded by the
+  handler with a single-variable filter, validated so a missing secret fails
+  loudly.
+- `Network=host` exporters scraped by containerized Prometheus must bind the
+  host's LAN address (containers cannot reach host loopback) PLUS `127.0.0.1`
+  for the homepage probe — two `--web.listen-address` flags, never the
+  wildcard.
+- alloy 1.x refuses the `ui` config block; the unit must pass
+  `--server.http.listen-addr=0.0.0.0:12345` before its port can be published.
+- GPU services: `AddDevice=` + `GroupAdd=keep-groups` (user-unit spawns carry
+  no supplementary groups).
+- When the last service leaves a compose file, the compose-validity seam
+  fails (requires ≥1 service): keep the file as comments and relax the seam
+  deliberately, or delete the file and fix the planner's compose-diff path —
+  decide with the owner, the deploy tooling reads it.
+
+## 3. Drift / crash-loop diagnosis (5 minutes, in order)
+
+Symptom: crash-looping or perpetually unhealthy container whose DEFINITION
+looks right; or a service quietly down for days because a restart policy kept
+something running. Cause: the running container was created from an OLDER
+definition — a compose/unit edit does not touch an existing container, and
+the restart policy faithfully resurrects it forever.
+
+1. **Who started it:** `podman inspect <name> --format 'unit={{index
+   .Config.Labels "PODMAN_SYSTEMD_UNIT"}} restart={{.HostConfig.RestartPolicy.Name}}
+   created={{.Created}}'`; compare `created` with when the definition changed
+   (`git log -p --follow <config>`). Created-before-the-edit is the signature.
+2. **What it is ACTUALLY running:** `{{.Config.CreateCommand}}` is the truth.
+   `{{.HostConfig.Devices}}` / `{{.HostConfig.GroupAdd}}` print empty/0 even
+   for containers started with `--device`/`--group-add` — a false negative
+   that sends you hunting the wrong root cause.
+3. **What the process says:** `podman logs <name> | grep -iE 'error|cannot|Runtime'`
+   (a capability error like `No HIP GPUs are available` beside a missing
+   device flag confirms drift). Count the damage: `{{.RestartCount}}` — one
+   instance reached 9,146 restart attempts on stale config.
+
+Quickest unblock: `podman compose up -d --force-recreate <svc>`. Durable fix
+= the §2 cutover. **Verify the capability, not the status:** `podman exec
+<name> python3 -c "import torch; print(torch.cuda.is_available())"`;
+`ss -ltn | grep <port>` for the actual bind addresses. ComfyUI incident:
+2026-09-16, stale device-less container, 9,146 restarts. History repeats:
+2026-08-26 Caddy (no unit), 2026-09-11 five observability services.
+
+## 4. Podman 5.x quadlet v1 generator quirks (verified live 2026-08-28)
+
+- **Hard-fail keys** — the whole unit silently fails to generate
+  (`systemctl --user cat <unit>` shows "could not be found"): any
+  dash-prefixed `[Container]` option (`--shm-size=…` → `unsupported key …
+  in group 'Container'`); `Command=` is unsupported, use `Exec=`.
+- **HealthCmd space-splitting:** the v1 generator splits `HealthCmd` on
+  spaces into the healthcheck argv; a quoted probe
+  `/bin/sh -c "pg_isready …"` lands as argv token `"pg_isready` →
+  command-not-found → container reads **unhealthy while the DB is up**.
+  HealthCmd must contain only space-free tokens
+  (`HealthCmd=pg_isready -d authentik -U authentik`). If health reads
+  unhealthy but logs look fine, compare the baked
+  `{{.Config.Healthcheck.Test}}` array against the intended command before
+  believing the status.
+- **Probe a new key before shipping:** drop a throwaway `probe.container`,
+  `daemon-reload`, `journalctl --user --since "1 min ago" -o cat | grep -i
+  "unsupported key"`, `systemctl --user cat probe.service` (shows the
+  generated ExecStart verbatim), then remove + reload.
+
+## 5. Rootless Quadlet log retrieval
+
+Step 0 — establish ownership: `systemctl --user status <name>.service`
+(works ⇒ rootless user unit); `loginctl show-user <user> -p Linger`. If
+`--user` knows the unit, EVERY command runs as that user with NO sudo.
+
+| Need | Command |
+|---|---|
+| Current instance tail/follow | `podman logs --tail 200 / -f <name>` |
+| Across restarts/recreates | `journalctl --user -u <name>.service -n 200 --no-pager` |
+| Time-bounded / crash-only | `--since "14:30"` / `-p warning` |
+| Kernel-side evidence (OOM, SVM) | `sudo journalctl -k --since …` |
+
+Quadlet units commonly run `--replace --rm`: **`podman logs` is destroyed on
+recreation** — the journal is the only durable record. Traps:
+`sudo podman logs` → "no such container" (root's store does not contain
+rootless containers; same for `sudo podman ps`); a system-level
+`journalctl -u <name>` returns nothing for a user unit; during first-start
+weight downloads health reports unhealthy BY DESIGN (check the payload file
+is growing). Verbose vendor switches are a config change — through git and
+deploy.sh, never a live env poke.
+
+## 6. Compose bring-up with ground truth (rootless hosts)
+
+1. **Audit reality before pressing buttons:** `podman ps -a` runs per-user —
+   check every namespace (`sudo podman ps -a`) and native units
+   (`systemctl list-units`) before declaring a service absent. Compare
+   compose mount sources against the real filesystem — compose files name
+   phantom paths (e.g. `/mnt/media/*` vs the real LUKS array at
+   `/mnt/entropy`). Fix compose through git first, deploy second.
+2. **Pre-create bind-mount sources** (`sudo mkdir -p` + `chown 1000:1000`)
+   or creation fails `statfs …: no such file or directory`.
+3. **Staged bring-up:** infra first (`up -d mariadb redis`), poll
+   `{{.State.Health.Status}}`, then the rest. Fresh containers answer
+   000/refused for the first seconds — poll.
+4. **Never guess image tags:** ghcr.io answers 403 on the bearer-token
+   request for nonexistent/private repos ("manifest unknown"/token len 0 =
+   repo or tag does not exist, not auth). Query the registry API
+   (`ghcr.io/token?scope=repository:<ns>/<repo>:pull` →
+   `/v2/<ns>/<repo>/tags/list`).
+5. **Read the interface from inside the image** before guessing flags/envs:
+   `podman exec <ctr> cat /z/cpp.sh` etc. HTTP 500 with a styled error page
+   means RUNNING-but-misconfigured, not down. Copyparty ac ground truth:
+   entrypoint `/bin/ash /z/cpp.sh -c /z/initcfg`; accounts/volumes are ARGV
+   flags (`-a USER:PASS`, `-v SRC:DST:PERMS`), not envs — and with none
+   configured it serves cwd read/write for everyone (an unauthenticated
+   world-writable share).
+6. Remote-shell quoting: zsh eats bare `===`/`?` tokens; a zero-match from a
+   mangled pattern is not absence — re-verify the target exists.
+7. `user: "1000:1000"` SIGSEGV at boot = unmapped uid in the rootless userns
+   (host 1000 == container 0): drop the user line, run as container-root.
+
+## Related
+
+- [[Deployment Flow]] — the deployer that installs these units
+- [[Stale Signal Audit]] — the audit that catches drift's victims
+- [[Container Network Namespaces]] — reachability between containers
+- [[Dankclaw/ComfyUI]] — the 9,146-restart incident
+- [[Infrastructure/Observability]] — same class, twice

serverAppConfigs-docs/Architecture/Stale Signal Audit.md

@@ -0,0 +1,144 @@
+---
+type: reference
+title: Stale Signal Audit
+description: The audit battery for surfaces that rot without failing — unreachable probes, vacuous assertions, invalid checkers, gates that never run, and prose invalidated by a change. Four passes found real defects every time.
+tags: [testing, observability, audit, signals, gates]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Stale Signal Audit
+
+Run after any change and before trusting a green gate. The productive question
+is **which signal here cannot possibly be correct?** Every pass on this repo
+(4 of 4) found real defects — including one in the rule file every future
+agent reads first.
+
+## The surfaces, and what each vantage point can reach
+
+| Surface | Producer's vantage point | Can reach |
+|---|---|---|
+| Homepage status cards | the `homepage-status` timer on **Dankclaw** | Dankclaw's loopback, Dankclaw container names, **published** LAN addresses |
+| Prometheus scrape jobs | the `prometheus` container on Dankclaw | container names on `dankclaw_default`, LAN addresses, that host's loopback only |
+| bats / e2e suites | the dev box | the repo, plus the live fleet over HTTPS/LAN |
+
+Anything else — another host's loopback, a container-internal port never
+published — is unreachable no matter how healthy the service is. The failure
+mode is always the same shape: a signal that **cannot** succeed by
+construction, which teaches the operator to ignore the surface it lives on.
+
+## 1. Permanently-red probes (reachability)
+
+For every monitor/probe/alert target ask: *can this checker physically reach
+that endpoint?* Seen classes: a card probing another host's container-internal
+port; a "probe" job scraping `/` (HTML → `unsupported Content-Type`); a
+`/metrics` that is disabled (Gitea 404).
+
+- A scrape cannot express "is this HTTP service answering?" — the right probe
+  is the **blackbox exporter** in probe mode (which also gives guarded vhosts
+  a real uptime signal, since Prometheus cannot solve an Anubis challenge).
+- **Fix in the direction of honesty, never of silence:** unreachable target →
+  mark the homepage card `wip: true` with the reason in `description` (the
+  checker skips `wip`; `wip` is the existing convention for "real service,
+  not probeable"), or repoint the job at something reachable. NEVER open a
+  database port on the LAN to make a badge green, and never delete a scrape
+  job — an absent scrape and a healthy one are indistinguishable.
+- Alert rules must not depend on always-red signals — scope them to services
+  that can actually be healthy (`LlmEndpointDown` is scoped to `frank-*` for
+  exactly that reason).
+- A target's failure STRING can prove a fix: a `/metrics` scrape flipping from
+  `connection refused` to `unsupported Content-Type "text/html"` means the
+  service came up and only the scrape is wrong. Read the error, not just
+  up/down.
+
+## 2. Vacuous assertions (highest severity — the gate is already dead)
+
+A seam that greps a file passes *by absence* once the file is retired:
+`grep` exits 2 on a missing path and `!` inverts it.
+
+```bash
+grep -rn 'grep.*"\$REPO_ROOT/[^"]*"' tests/*.bats | head -50
+[ -f "$path" ] || echo "MISSING: a seam may be grepping this"
+```
+
+Fix shape — one place asserts the invariant, each local site keeps a live
+check:
+
+```bats
+@test "no <host> compose file competes with the quadlet units" {
+    local cf="$REPO_ROOT/<host>/compose.yaml"
+    [ ! -f "$cf" ] || { echo "$cf is back: every container is unit-managed"; return 1; }
+}
+# at the per-service site: still fails if the entry returns
+[ ! -f "$cf" ] || ! grep -qE '^\s+<svc>:' "$cf"
+```
+
+**Prove teeth:** recreate the offending file
+(`printf 'services:\n  <svc>:\n    image: x\n' > <path>`), run the seams,
+confirm RED, then move the file to trash. A re-armed check you have never
+seen fail is still unverified.
+
+## 3. Gates that never run
+
+Playwright specs in `tests/e2e/` are matched by **number prefix** in
+`playwright.config.ts`; an unregistered spec silently never runs and the
+suite count stays the same. Count tests before/after;
+`tests/e2e_scaffold.bats` fails on an unregistered spec by design.
+
+## 4. Validate the checker before acting on it
+
+A checker's output is a hypothesis about another tool's semantics. Modelling
+Obsidian with naive text matching produced 45 "broken links" of which 3 were
+real — the rest were case-sensitivity (a `qBittorrent` link to
+`qbittorrent.md`), path-qualified resolution (a path-qualified
+`folder/Note` link resolves by path SUFFIX, not
+stem), and `%20` URL-encoding in markdown links. Fix the checker first,
+re-derive the report, only then edit documents. The result lives in
+`scripts/check-vault-links.py` with those three rules in its docstring,
+gated by the `no unresolved links in the docs vault` seam. Same class: a
+`find … | head -1` selector that reads the wrong file; a truncated listing
+that makes you "fix" a file that exists.
+
+## 5. Prose invalidated by your own change
+
+After retiring a file or adding a feature, grep for EVERY reference, not just
+code:
+
+```bash
+grep -rn "<retired-file>" --include="*" . | grep -v '^./.git/'
+grep -rn "no <feature>\|none —\|unconfigured" --include="*.md" --include="*.container" --include="*.yml" .
+```
+
+This catches the ones that bite: AGENTS.md said ComfyUI runs CPU-only long
+after it was restored with GPU access; a unit said "there are no alerting
+rules" hours after five were added; a docs page documented three scrape jobs
+as working when they could never report up. **Clean cutover includes prose** —
+docs, unit docstrings, and the repo's rule file (the most expensive place to
+be stale).
+
+## Order and verification
+
+1. Enumerate the surfaces (status board, metrics targets, alert rules, test
+   gates, docs/rule files, dashboard panels).
+2. For each, ask what the signal CAN express, not what it says.
+3. Validate any checker before editing the things it flags.
+4. Fix at the root; never weaken an assertion to make it pass.
+5. Gate the class so it cannot return (a seam), then prove the seam fails on
+   the defect.
+
+Verify: `./scripts/check-vault-links.py` (exit 1 on breaks); query each
+Prometheus panel expression against the API (0 series on a panel that should
+show data is a defect; 0 `ALERTS` when nothing fires is not); read the
+deployed `/srv/homepage/status.json` and triage every `state != up` entry
+against reality; query `/api/v1/targets?state=active` and print
+`health != "up"` with `lastError`; `./tests/run.sh` → `ALL TESTS PASSED`.
+Report per finding: the surface, why the signal could not have been correct,
+the fix, and the red/green evidence — flag any regression you introduced
+yourself in the same breath (one was found this way: a tightened bind killed
+a reverse-proxied vhost while every gate stayed green, because the spec
+covering that hostname did not exist).
+
+## Related
+
+- [[Testing Policy]] — the seam inventory this audit walks
+- [[Observability]] — the scrape/probe surfaces it audits
+- [[Quadlet Lifecycle]] — the ownership labels the probes read

serverAppConfigs-docs/Dankclaw/Hermes.md

@@ -81,6 +81,178 @@ The wiring lives in the live `/data/config.yaml` under `mcp_servers.chrome-devto
 The npx package cache lives in the container layer, so a recreation (not just a
 restart) pays one first-call npm download; subsequent calls are cached.
 
+## Config-home drift ("cannot connect" to models that answer curl)
+
+`podman exec <container> printenv HERMES_HOME` for each hermes container. If
+the webui reads the image default (`/opt/data`) while the gateway reads
+`/data`, the webui runs the image-default template config (`provider: auto` →
+openrouter.ai with no key = connect failure).
+
+The recreation trap is the same one the volumes table shows: image `VOLUME`
+points get a fresh anonymous volume on every container recreation
+(`podman run --replace --rm`) — edits inside such a mount are lost on
+restart/redeploy. Compare volume ids before/after a restart to confirm.
+
+Fix: pin `Environment=HERMES_HOME=/data` in the quadlet unit (shared state
+root), deploy service-scoped. Verify with the app's own code, not guesses:
+
+```python
+from api import config as C; C.resolve_model_provider("<model-id>")
+# -> (model_id, provider, base_url); bare ids map to the owning provider by
+#    scanning the providers: dict models allow-list; @provider:model forces one.
+hermes_cli.providers.resolve_user_provider(name, cfg["providers"])
+# -> ProviderDef with real base_url; accepts api/url/base_url keys.
+```
+
+Named providers (`providers:` dict) are a catalog + connection source;
+`custom:<slug>` connections resolve only via `custom_providers[]` entries. If
+neither holds the provider name, the request has no endpoint and fails to
+connect. Probe the endpoint with the **exact served model id** — the retired
+FLM/NPU endpoints routed only by exact registry tag and answered HTTP 200
+with `{"error":"Failed to load <tag> model!"}` for unknown tags; status-code
+checks miss this class entirely.
+
+## Live config edits
+
+Live file: `/home/dank/Containers/hermes/data/config.yaml` (dir `drwx------`
+uid 109999; all reads/edits need `sudo`). The repo file
+`dankclaw/hermes/config.yaml.example` is a sanitized backup and **can lag
+live — always diff the live file first; never treat the example as truth**.
+
+1. **Backup first:** `sudo cp -a …/config.yaml …/config.yaml.bak-YYYYMMDD`
+   (in-dir `.AGENTTRASH/` also exists for removals).
+2. **Apply** with surgical `sudo sed -i` or full-file write; YAML comments in
+   the live file are tool-generated — preserve them.
+3. **Sweep stale references** across the whole data dir:
+   `sudo grep -rln "<old-id>" /home/dank/Containers/hermes/data
+   --include="*.yaml" --include="*.json"`. Session files/MEMORY.md are
+   historical — leave them.
+4. **Derived caches:** `webui/models_cache.json` is keyed by a config
+   fingerprint (`_source_fingerprint.config_yaml` mtime/size); after edits it
+   is stale. Let it regenerate lazily or trash it — never hand-edit.
+5. **Restart the model-consuming units** (they parse config at startup):
+   `systemctl --user restart hermes-webui.service hermes.service`.
+   `hermes-dashboard.service` is unaffected by routing changes.
+6. **Verify behaviorally**: one tiny completion per endpoint with the exact
+   served id — POST `/v1/chat/completions` and read the body, never just the
+   status code (endpoint names rotate; see the LLM knowledge base).
+7. **Sync the repo backup** (`dankclaw/hermes/config.yaml.example`) to the
+   live shape, redacting secrets (`password_hash`, phone account stay
+   live-only), then `./tests/run.sh` before commit.
+
+`provider: custom` + `base_url` entries route by base_url and send the model
+id verbatim; named providers resolve via their `api:` URL. **Renaming a
+served alias anywhere (serve scripts, env ALIAS) MUST propagate to this
+config in the same change** or Hermes requests 404 on nonexistent models —
+this exact gap caused the 2026-09-05 outage.
+
+## Auxiliary routing against local backends
+
+When a Hermes feature that calls an auxiliary task (e.g. title generation)
+fails against a local endpoint, the usual cause is a **request shape the
+local runtime does not implement**, not a routing mistake.
+
+Diagnose with the real call path — run the actual function inside the
+container with the app's own constants (the probe cannot drift from
+production):
+
+```bash
+base64 -w0 /tmp/probe.py | ssh dank@192.168.0.143 \
+  'base64 -d > /tmp/probe.py && systemctl --user restart hermes.service && sleep 20 && \
+   podman cp /tmp/probe.py systemd-hermes:/tmp/probe.py && \
+   podman exec -i systemd-hermes /opt/hermes/.venv/bin/python /tmp/probe.py'
+```
+
+Copy the file AFTER any restart (`--replace --rm` discards the container
+layer). The CLI at `/opt/hermes/hermes` uses system python and dies
+`No module named 'yaml'`; use `/opt/hermes/.venv/bin/python`.
+
+```python
+import sys; sys.path.insert(0, "/opt/hermes")
+from agent.title_generator import _TITLE_PROMPT_TEMPLATE, _TITLE_RESPONSE_FORMAT, _clean_title, _extract_title_text
+from agent.auxiliary_client import call_llm
+resp = call_llm(task="title_generation", messages=[...], max_tokens=64,
+                temperature=0.3, extra_body={"response_format": _TITLE_RESPONSE_FORMAT})
+```
+
+What actually breaks, and the fix:
+
+- **`response_format` (json_object / json_schema) is a hard 400 on every
+  local backend here** — neither FreeToken nor halogen implements constrained
+  decoding. Config cannot override it: `call_llm` merges the task's
+  `extra_body` first and then `update()`s the caller's, so the hardcoded one
+  wins. Only temperature and max_tokens have reactive retry paths.
+- **Fix: move that one task to the Responses transport**, which never
+  forwards `response_format`:
+
+  ```yaml
+  auxiliary:
+    title_generation:
+      provider: dankserver-small        # names track the live config; diff first
+      model: frank-small
+      api_mode: codex_responses        # real per-task key
+      extra_body:
+        reasoning:
+          effort: none                 # mandatory, see below
+  ```
+
+- **Disable thinking for any cap-limited task.** The adapter omits reasoning
+  only when config says `reasoning.enabled: false`. The
+  `reasoning_effort: none` *shorthand* produces exactly that and leaves the
+  server default (thinking ON) — the 64-token title budget gets spent on the
+  think: empty content, `finish_reason: length`. The explicit
+  `extra_body.reasoning.effort: none` instead sends
+  `{"effort":"none","summary":"auto"}`, which the local `/v1/responses`
+  honors.
+- Both local runtimes also answer `/v1/responses` and `/v1/messages` — no
+  proxy needed for the alternate transport.
+
+Verify and land: re-run the probe with **no** `api_mode`/reasoning kwargs —
+it must pass from config alone (that is the production path); keep the live
+file and the repo example in step; `./tests/run.sh`, commit, push — a Hermes
+config change is docs+example only, no service deploy needed.
+
+Gotchas: `auxiliary.<task>.enabled: false` is checked BEFORE
+`apply_instant_title`, so disabling a broken title task also removes the
+derived title — it does not degrade gracefully the way a failure does. A
+local call can also fail at the model level: a 200 with an error-text body,
+or a chatty non-title answer. Always read the raw content, not the status.
+
+## Webui password rotation ("Hermex")
+
+Auth contract (verified live 2026-08-31): `POST /api/auth/login` with JSON
+`{"password":"…"}` → 200 + signed HttpOnly `hermes_session` cookie (24 h TTL)
+on match, 401 on miss. No username field. `/health` answers 200
+unauthenticated and `/` answers 302 — **never verify auth via those**; probe
+login with a deliberately wrong password (expect 401) and the real one
+(expect 200). Details on the webui surface: [[Hermes-WebUI]].
+
+1. Run the whole probe as a heredoc'd bash script
+   (`ssh -o BatchMode=yes dankclaw 'bash -s'`) — the host's zsh mangles
+   inline `${…}` even inside single-quoted ssh args; never interpolate curl
+   `-w %{http_code}` inline.
+2. Update BOTH env files with
+   `sed -i "s/^HERMES_WEBUI_PASSWORD=.*/HERMES_WEBUI_PASSWORD=<new>/"`:
+   - `/home/dank/serverAppConfigs/.env` ([Service] EnvironmentFile)
+   - `/home/dank/serverAppConfigs/dankclaw/.env` — **the effective one**: the
+     quadlet's `[Container] EnvironmentFile=` is what the container reads at
+     create; rotating only the root file leaves the running container serving
+     the old secret.
+3. `systemctl --user restart hermes-webui.service` (unit recreation re-reads
+   env; no deploy needed if the unit file itself didn't change).
+4. Verify wrong-pw → 401, right-pw → 200, from host loopback **and** a LAN
+   client at `http://192.168.0.143:8787` (webui publishes on the LAN
+   interface since commit `4443897`; off-network ingress is SSH tunnel only —
+   user `tunnel`, port **22**, key-only auth, forward remote end
+   localhost:8787).
+
+Gotchas: `sudo podman exec` fails exit 125 here — root's namespace does not
+see rootless containers; run `podman exec hermes-webui printenv
+HERMES_WEBUI_PASSWORD` unprivileged (the ssh user owns the env files, mode
+600). If rotating because a secret leaked into transcripts/logs, rotate in
+place via a script that fetches the live value from the running container
+(`printenv`) instead of transcribing it.
+
 ## See Also
 
 - [[Hermes-Dashboard]] — the LAN browser management UI

serverAppConfigs-docs/Dankclaw/OpenWebUI.md

@@ -64,6 +64,49 @@ The Ollama connection is disabled (`ollama.enable=false`).
 > repository. Model swaps, context sizing, and spec tuning happen there. See
 > the [LLM knowledge base](../LLM/index.md).
 
+## Container introspection (DB and backend python)
+
+- Working dir for imports: `/app/backend`; package name `open_webui`
+  (underscore), NOT `openwebui`. `os.environ["WEBUI_SECRET_KEY"]` MUST be set
+  BEFORE importing any `open_webui` module (import-time hard requirement);
+  the secret lives at `/app/backend/.webui_secret_key`.
+- **The `sqlite3` CLI is NOT in the container** — use Python's `sqlite3`
+  module (`podman exec open-webui python3 -c "import sqlite3;
+  conn = sqlite3.connect('/app/backend/data/webui.db'); …"`). For config
+  keys use parameterized LIKE queries (avoids shell quoting).
+- Backend python: `podman exec -w /app/backend open-webui python3 -c "…"` —
+  `Models.get_all_models()` is a coroutine (await it); model objects are
+  pydantic `ModelModel` instances (attribute access, not `.get()`), with NO
+  `status`/`enabled` attribute — use `is_active` from the DB.
+- **API auth for verification:** `create_token` lives in
+  `open_webui.utils.auth` (NOT `open_webui.utils.utils`);
+  `get_verified_user` requires an `id` claim (not just `sub`);
+  `SESSION_SECRET = WEBUI_SECRET_KEY`. Mint an admin token inside the
+  container:
+
+  ```
+  podman exec -i open-webui sh -c 'export WEBUI_SECRET_KEY=$(cat /app/backend/.webui_secret_key); \
+    python -c "from open_webui.utils.auth import create_token; \
+    print(create_token({\"id\":\"<UID>\",\"sub\":\"<UID>\",\"email\":\"…\",\"role\":\"admin\"}))"'
+  ```
+
+  (admin UID from the `user` table).
+
+## Verification battery
+
+1. `GET /api/v1/models/list?limit=100` → returns EXACTLY the intended
+   models. Response key is `items`, not `data`.
+2. `POST /openai/chat/completions` per model with a trivial prompt ("Reply
+   with exactly one word: pong") → each returns content `pong`. Reasoning
+   models need `max_tokens>=100` — the thinking block eats the budget.
+3. A 500 "Server Connection Error" → `podman logs open-webui` for
+   `ClientConnectorError`: a network-reachability problem (host LAN IP vs
+   `host.containers.internal`), not a model problem.
+
+**Durability:** the model table + connections are DB-only (not git-managed);
+a fresh data volume loses them. For reproducibility, add an idempotent
+first-start seed script or document the setup in the service doc.
+
 ## Web search
 
 Chat web search is served by [[SearXNG]] on this host, wired in the quadlet
@@ -147,6 +190,23 @@ The quadlet carries a built-in healthcheck (30s interval, 3 retries, 300s start
 
 `podman ps` shows the container as `unhealthy` if either fails. Probe manually at `http://dankclaw:3005/health`.
 
+## Image generation (ComfyUI)
+
+Image generation rides the local [[ComfyUI]] integration, not the chat
+picker — ComfyUI's models never appear in the chat model picker (that is
+chat LLMs only), so ComfyUI needs its own image-generation connection
+configured in OpenWebUI. ComfyUI itself runs GPU-enabled on Dankclaw and
+serves `:8188`.
+
+Endpoint: `POST /api/images/generations` (image `0.11.x`: upstream moved the
+route to `POST /api/v1/images/generations`; the `comfyui` engine is still
+upstream) with `Authorization: Bearer <token>` and body
+`{"model":"imagegen","prompt":"…","n":1,"size":"1024x1024"}`; the response
+carries base64 image data or a URL.
+
+Failure modes: connection refused ⇒ OpenWebUI or ComfyUI is down; timeouts
+on large sizes ⇒ check ComfyUI resources; 401 ⇒ the token is bad.
+
 ## Retired instances
 
 - The compose `open-webui` service (port 3000) was never run: 3000 is Gitea's HTTP port, so it could not bind.

serverAppConfigs-docs/Dankclaw/Swap Configuration.md

@@ -117,6 +117,18 @@ frank-large `health=200` throughout.
   the model is loaded. Trust the engine's `host memory left for everything
   else` line, not `free`.
 
+- **Filesystem prerequisite (other hosts).** ext4 and xfs take swapfiles
+  directly. btrfs needs `chattr +C` (NOCOW), no compression, and no
+  snapshots in the swapfile path — check `df -T` before planning, or
+  `mkswap` will reject the file.
+- **Renaming an *active* swapfile works (inode-based) but leaves a stale
+  path in `/proc/swaps`** — swap it off first, then rename, then `swapon`
+  (which is exactly what the step-3 consolidation does). Migration is
+  observable while it runs: in the 2026-09-12 run `swapoff` of 4 GiB took
+  65 s and used-RAM rose only ~3 GiB (pages touched during migration), the
+  rest relocating on disk. If used-RAM climbs toward the whole swapped
+  amount, let it finish rather than aborting.
+
 ## Trash
 
 The previous 8 GiB `/swap.img` (dated 2026-07-21) is staged, not deleted, at

serverAppConfigs-docs/Desktop/Noctalia and Niri.md

@@ -0,0 +1,303 @@
+---
+type: reference
+title: Noctalia and Niri
+description: The noctalia v5 desktop shell on niri (workstation + blankpad) — config layers, remote IPC over ssh, custom-palette recolor, offline template rendering, and output rotation/pinning.
+tags: [desktop, noctalia, niri, blankpad, workstation, ghostty]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Noctalia and Niri
+
+Noctalia (`/usr/bin/noctalia`) is the desktop shell on niri, running on both
+the workstation and blankpad. Since the 2026-09-10 unification both run the
+**bare `noctalia` v5 binary** (niri `spawn-at-startup "noctalia"`); the old
+Quickshell-hosted v4 architecture (`qs -c noctalia-shell`, JSON config) is
+retired on both hosts.
+
+## Config layers
+
+- Hand-edit declarative layer: `~/.config/noctalia/<name>.toml` (e.g.
+  `config.toml`). Main config: `~/.config/noctalia/config.toml`; the niri
+  config includes `noctalia.kdl` for display settings.
+- Runtime state file `~/.local/state/noctalia/settings.toml` loads LAST and
+  **overrides declarative values for the same keys — never hand-edit it**;
+  UI/IPC writes land there.
+- Main reference: `noctalia-reference` facts — IPC `noctalia msg <command>`
+  replaced the `qs -c noctalia-shell ipc call` wrapper; app templates live at
+  `/usr/share/noctalia/assets/templates/` (gtk, ghostty, alacritty, hyprland,
+  sway, wezterm, helix, btop, starship, qt, kde, cava, scroll, mango,
+  umbriel, foot, emacs, kitty); palette generation `noctalia theme [image]`.
+
+## Remote IPC over ssh (the D-Bus trap)
+
+`noctalia msg <command>` resolves the running instance via the D-Bus session
+bus. sshd-spawned non-login shells lack `DBUS_SESSION_BUS_ADDRESS`, so plain
+ssh exec prints `error: noctalia is not running` even when the shell IS
+running — do not conclude "shell dead" from this error. Working remote
+procedure (runs in the systemd user-manager context where the bus address
+exists):
+
+```
+systemd-run --user --pipe --wait noctalia msg <command> [args]
+```
+
+Verified: `msg status` returns live JSON state; unknown panel ids return an
+error listing all valid panel ids (clipboard, control-center, launcher,
+polkit, session, setup-wizard, test, tray-drawer, wallpaper).
+
+Verified v5 msg verb grammar (keybinds-safe): `panel-toggle <id>`,
+`session lock|suspend|lock-and-suspend|logout|reboot|shutdown`,
+`volume-up [step]`, `volume-down`, `volume-mute`, `mic-mute`,
+`media next|previous|playPause` (graceful no-op, rc 0, with no active
+player), `brightness-up [target] [step]`, `brightness-down`,
+`settings-open [context]`, `settings-toggle [context]`, `config-reload`,
+`wallpaper-get [connector]`, `wallpaper-set <path>`, `wallpaper-next`,
+`color-scheme-get|set`, `theme-mode-get`, `templates-apply`,
+`screenshot-fullscreen`.
+
+**Wallpaper directory semantics:** empty/unset `[wallpaper] directory`
+resolves to XDG Pictures via user-dirs.dirs; an explicit directory key
+overrides. Fleet default pattern: point `directory` at the repo-tracked
+wallpapers dir so git-tracked wallpapers serve the shell without copying
+files. After writing config.toml over ssh the file watcher hot-reloads
+automatically; `config-reload` via systemd-run forces it immediately. Note:
+the *current* wallpaper stays the built-in default until the next rotation
+or an explicit `wallpaper-next`; `wallpaper-get` then reporting a path
+inside the configured directory is the confirmation the config took effect.
+
+**Lock chain (unified):** swayidle-lock.service ExecStart on both hosts:
+`/usr/bin/swayidle -w timeout 300 '/usr/bin/noctalia msg session lock'
+before-sleep '…' after-resume '…'`. `msg` resolves via the session bus, so
+no WAYLAND_DISPLAY pinning is needed in unit context.
+
+## Custom-palette recolor (without restarting the session)
+
+Why custom, never the community cache:
+`~/.local/state/noctalia/community-palettes/*.json` is a **sync cache** —
+`.catalog/palettes.json` carries an `md5` per palette and noctalia re-fetches
+from `https://api.noctalia.dev/palettes`, so edits there get reverted.
+Custom palettes live at `~/.config/noctalia/palettes/<Name>.json`
+(`NOCTALIA_CONFIG_HOME` then `XDG_CONFIG_HOME` honored) and are never
+synced. Missing/invalid file → silent fallback to the builtin palette.
+
+Palette JSON shape = the community file's shape: `dark`/`light` maps of
+`mPrimary`, `mOnPrimary`, `mSecondary`, `mTertiary`, `mError`, `mSurface`,
+`mSurfaceVariant`, `mOutline`, `mShadow`, `mHover` plus a `terminal` block
+(`foreground`, `background`, `normal.{black..white}`, `bright.{…}`,
+`cursor`, `cursorText`, `selectionFg`, `selectionBg`). NOTE: `noctalia theme
+-o` emits M3 snake_case tokens (`on_primary_container`, …) — that is the
+*generator* format, not this one. Copy an existing community JSON and mutate
+with `jq`.
+
+Procedure:
+
+1. Read the live scheme first — never assume: `noctalia msg color-scheme-get`
+   (e.g. `community Gruber Darker`); `noctalia msg theme-mode-get`.
+2. Back up settings: `cp -a ~/.local/state/noctalia/settings.toml
+   /tmp/noctalia-settings-backup-$(date +%s).toml`.
+3. Build the palette from the current one (jq keeps structure):
+
+   ```sh
+   mkdir -p ~/.config/noctalia/palettes
+   jq '.dark.mPrimary = "#f43841"
+       | .dark.terminal.normal.yellow = "#f43841"
+       | .dark.terminal.bright.yellow = "#ff4f58"
+       | .dark.terminal.foreground   = "#f43841"' \
+      "$HOME/.local/state/noctalia/community-palettes/Gruber%20Darker.json" \
+      > ~/.config/noctalia/palettes/"Gruber Red.json"
+   ```
+
+   Prefer a hex already present in the palette (its own `mError` red) so the
+   theme stays internally consistent. Set BOTH `normal` and `bright`
+   variants, and both `dark` and `light` if the user toggles modes.
+4. Switch (persists `theme.source = custom` + `theme.custom_palette`):
+   `noctalia msg color-scheme-set custom "Gruber Red"` → `ok`; confirm with
+   `color-scheme-get`.
+5. Re-render every configured template: `noctalia msg templates-apply`.
+   Regenerates `~/.config/ghostty/themes/noctalia`,
+   `~/.config/gtk-{3,4}.0/noctalia.css`, kcolorscheme, foot, wezterm,
+   starship, etc. Symlinked-into-dotfiles paths show up as modified files
+   worth committing.
+6. Reload surfaces: compositor — `niri msg action load-config-file`
+   (`reload-config` is NOT a valid niri subcommand); ghostty — see the trap
+   below.
+
+Verification: `ghostty +show-config | grep -E '^(foreground|palette)'`
+(proves the theme file resolves — note it prints only the delta from
+defaults; cross-check with `ghostty +show-config --default` before
+concluding a key is unset); `noctalia config validate` → `✓ Config is
+valid`; visual via `noctalia msg screenshot-fullscreen` (saves to
+`~/Pictures/screenshot_<ts>.png` and ALSO overwrites the clipboard with the
+PNG — warn the user); `grep -c <old-hex> ~/.config/noctalia/palettes/*.json`
+→ 0 for no leftovers.
+
+Track it in the dotfiles repo (`~/.omp`, the command-center —
+`scripts/deploy.sh` is a stow-equivalent over `dotfiles/<pkg>/` trees):
+mirror the shape for the new palette so `git pull` + `deploy.sh` is the
+whole propagation story — move the JSON into
+`~/.omp/dotfiles/noctalia/.config/noctalia/palettes/` and symlink it back
+into `~/.config/noctalia/palettes/`; re-set the custom palette (re-reads
+through the symlink). Do the link by hand for one file instead of running
+`deploy.sh` — sync mode quarantines and replaces every divergent live file.
+
+Traps:
+
+- **NEVER `kill -USR1` ghostty to reload.** It does not reload config on
+  this fleet — it kills/respawns the terminal (took out Frank's window
+  2026-09-13). Correct: the default keybind `ctrl+shift+,` = `reload_config`
+  (per `ghostty +show-config --default`). Existing windows keep old colors
+  until reloaded; new windows inherit immediately.
+- Palette hex values are NOT a proxy for what the user sees: in
+  `Gruber Darker` the *accent* (`mPrimary`) is yellow `#ffdd33`, so "change
+  the yellow" means the accent too, not just terminal index 3.
+- Saturating terminal text makes ANSI 1/9 indistinguishable from default
+  text (grep highlights flatten) — flag it and offer a lighter red.
+- Contrast sanity: `#f43841` on `#181818` ≈ 4.6:1 (dark-on-color beats
+  white-on-color here, ~3.9:1 — keep `mOnPrimary` dark).
+- Revert = `noctalia msg color-scheme-set community "<original name>"`; the
+  community cache is untouched by the whole flow.
+- Generated theme files symlinked into the dotfiles repo (`gtk-3.0`/`gtk-4.0`
+  `noctalia.css`, ghostty theme) show up as tracked modifications after
+  `templates-apply`; commit them together with the new palette JSON so every
+  machine gets the same recolor.
+
+## Offline template render (when templates-apply never ran)
+
+Failure signature: niri `failed to read included config from
+~/.config/niri/noctalia.kdl: No such file or directory` — the desktop runs
+but with DEFAULT config (niri 26.04 falls back, does not exit): no custom
+keybinds/rules, no noctalia autostart, default output placement. ghostty:
+`theme "noctalia" not found, tried ~/.config/ghostty/themes/noctalia`.
+alacritty: import of `~/.config/alacritty/themes/noctalia.toml` missing.
+
+Root cause pattern: upstream dotfiles commits theme switches (niri include,
+ghostty `theme = noctalia`, alacritty import) that depend on noctalia
+TEMPLATE-GENERATED files. Those files are per-host outputs (rendered from
+the local palette) and are intentionally NOT in the repo; on any host where
+`templates-apply` never ran, they don't exist.
+
+Fix (offline, works while a session runs) — built-in templates live at
+`/usr/share/noctalia/assets/templates/` (declared in `builtin.toml`); render
+each needed one with the live palette:
+
+```bash
+# niri include (plain host-local file; do NOT symlink into repo — regenerated per host)
+noctalia theme --theme-json ~/.config/noctalia/colors.json \
+  --render "/usr/share/noctalia/assets/templates/niri/niri.kdl:$HOME/.config/niri/noctalia.kdl"
+mkdir -p ~/.config/ghostty/themes && \
+noctalia theme --theme-json ~/.config/noctalia/colors.json \
+  --render "/usr/share/noctalia/assets/templates/ghostty/ghostty:$HOME/.config/ghostty/themes/noctalia"
+mkdir -p ~/.config/alacritty/themes && \
+noctalia theme --theme-json ~/.config/noctalia/colors.json \
+  --render "/usr/share/noctalia/assets/templates/alacritty/alacritty.toml:$HOME/.config/alacritty/themes/noctalia.toml"
+```
+
+All template tokens (primary, surface, error, primary_container, …) resolve
+from the flat 16-key colors.json; container tokens are derived, never blank.
+
+Verify: `niri validate` → rc 0; `ghostty +validate-config` → rc 0; next
+login journal shows `loaded config from config.kdl` with NO parse WARN,
+custom output placement (`niri msg outputs` vs display.kdl), noctalia
+autostarted (`pgrep noctalia`).
+
+Traps: **`noctalia msg templates-apply` on 5.0.1 returns "ok" but applies
+NOTHING** (this build's opt-in config schema does not act on it) — do not
+trust its exit code, use the explicit `--render` calls above. Niri config
+errors at boot are NOT fatal — the compositor keeps running on defaults,
+which reads as "some settings silently stopped working", not a crash.
+Regenerate after any palette change; the rendered files are static
+snapshots.
+
+## Niri output rotation and pinning
+
+Verified on blankpad, niri 26.04 (two PM1601 externals + eDP-1 laptop panel).
+
+**Ground truth first:** `niri msg outputs` (connector names, current mode,
+scale, transform, logical pos/size) and `niri --version`. Output **names are
+connector names** (`eDP-1`, `DP-2`, `HDMI-A-1`), not EDID model strings —
+the EDID strings printed before them are decoration. On blankpad the live
+config splits via `include "./cfg/*.kdl"` from `~/.config/niri/config.kdl`,
+and `cfg/display.kdl` is a **symlink into the dotfiles repo** — edit the
+repo path (that is what gets committed); check with `ls -ld`.
+
+**Transform direction is counter-clockwise** (`niri msg output <NAME>
+transform --help`): `90` = content rotated 90° CCW ⇒ the panel's **top
+bezel sits on the LEFT**; `270` = top bezel RIGHT; `180` = upside down;
+`flipped-90`/`flipped-270` = mirrored variants. `90` vs `270` is a
+physical-stand fact no tool can read (portable monitors expose no
+orientation in EDID/MCCS) — pick one, apply, and hand the user the one-liner
+to flip; don't stall on the question.
+
+Two traps when editing the config:
+
+1. **An `output "X" { … }` block resets every property you don't restate to
+   its default.** Pinning position on `eDP-1` without repeating `scale 1.25`
+   silently drops the laptop panel to scale 1.0. Restate `scale` (and
+   `mode`) in every block you touch.
+2. **Niri re-runs auto-layout for outputs with no pinned position.** Pinning
+   only the externals shoved `eDP-1` from `0,0` to `3936,0`. Pin every
+   output's position to keep the left-to-right order. Positions are logical
+   pixels at each output's own scale; x-offsets chain (eDP-1 logical width
+   1536 → DP-2 at 1536 → +1200 portrait width → HDMI-A-1 at 2736):
+
+```kdl
+output "DP-2"     { scale 1 transform "90" position x=1536 y=0 }
+output "HDMI-A-1" { scale 1 transform "90" position x=2736 y=0 }
+output "eDP-1"    { scale 1.25 position x=0 y=0 }   // MUST restate scale
+```
+
+Apply and verify:
+
+```bash
+niri validate -c ~/.config/niri/config.kdl      # exit 0
+niri msg action load-config-file                # live reload; NOT reload-config
+sleep 1; niri msg outputs | grep -E '^Output|Transform|Logical|Scale'
+```
+
+Success looks like `Logical size: 1200x1920` (swapped) plus
+`Transform: 90° counter-clockwise` on each rotated panel, positions matching
+the intended chain. One-off rotation without touching config (temporary):
+`niri msg output DP-2 transform 90`. Commit via the dotfiles repo
+(`~/.omp`, `main`; push to the local Gitea permitted): `git pull --ff-only`
+→ add the display.kdl → commit → push. Pitfalls: `reload-config` is not a
+niri subcommand; never SIGUSR1 a terminal to "reload" (niri config reloads
+do not touch ghostty themes); no man page for niri on this host — `niri
+<sub> --help` and `/usr/share/doc/niri/default-config.kdl` are the docs.
+
+## GPU RGB over I2C (EVGA iCX3)
+
+Turn RGB on/off (or read it back) on an EVGA FTW3 / iCX3-controller GPU
+(the 3090 on dankclaw) using the GPU's own nvidia i2c bus. No display
+engine, no X, no daemon — the mode is EEPROM-saved and survives reboot.
+
+- **Find the controller:** the nvidia i2c driver exposes the card's lighting
+  ports as `/dev/i2c-N`; on the 3090 FTW3 that is ports 1,5,6,7,8 →
+  i2c-1..i2c-5, and only i2c-1 carries a live slave (NVIDIA port 1). Probe
+  each bus. The iCX3 MCU answers at address **0x2D**. `/dev/i2c-*` are
+  `root:i2c 660` — run with `sudo` (dank has passwordless sudo).
+- **Protocol (raw I2C_RDWR, NOT the SMBus ioctls)** — write the register
+  byte first, then the payload:
+  - enable `0xB2` write `04 C6 EB EA 15` (required once before mode writes)
+  - mode `0xC0` write `09 m0 m1 m2 m3 .. .. .. .. SYNC`; m0..m3 are the four
+    zone modes, `MODE_OFF = 0x00`; OFF = `09 00 00 00 00 FF FF FF FF 00`
+  - save `0x90` write `04 9E EB 00 90` (commit to EEPROM — the
+    reboot-persistence step)
+  - fw `0xB1` block read (firmware version)
+  - Read the mode back: write reg `0xC0`, then read 16 bytes; bytes 1..4 are
+    the zone modes (0x00 = OFF), bytes 6..9 the per-zone LED counts.
+- **Verify persistence:** after a reboot, read the mode — all four zone
+  modes `0x00` means the EEPROM save held (confirmed 2026-09-19 post-reboot).
+- **Tool:** a ~90-line I2C_RDWR reader/writer lived at `/tmp/evga_raw.c`
+  (`evga_raw <bus> mode|off|save|enable|fw`) — `/tmp` is wiped by reboot, so
+  recreate it if needed; OpenRGB's `EVGAGPUv3Controller.cpp` carries the
+  same protocol plus the full colour API.
+- Gotchas: raw I2C_RDWR only (the iCX3 answers the raw form, not the SMBus
+  ioctls); the `enable` write must precede the first mode write in a
+  session; without the `save` write the mode is RAM-only and resets on power
+  cycle.
+
+## Related
+
+- [[Command Center Dotfiles]] — the stow-equivalent deploy this page links into
+- [[Hermes]] — the agent that drove the CDP daemon alongside this desktop
+- [[GGUF Compatibility]] — unrelated but shares the "read the file, not the name" lesson

serverAppConfigs-docs/FrankBook-Pro/ARR-Stack.md

@@ -5,3 +5,115 @@ description: "The *arr automation stack (Prowlarr, Sonarr, Radarr, Lidarr, Reada
 tags: [arr-stack, prowlarr, sonarr, radarr, lidarr, readarr, qbittorrent, frankbook-pro, dankserver]
 timestamp: 2026-09-04T00:00:00Z
 ---
+
+# *arr Stack
+
+The *arr automation stack runs on **FrankBook-Pro** (the Mac mini) as quadlets
+inside its podman machine; all egress rides the gluetun WireGuard netns. The
+per-service pages ([[Prowlarr]], [[Sonarr]], [[Radarr]], [[Lidarr]],
+[[Readarr]], [[qBittorrent]]) carry deployment facts; this page holds the
+cross-cutting API and auth-posture procedures.
+
+## API version split (memorize before any *arr HTTP call)
+
+- Sonarr/Radarr: `/api/v3/…`; Lidarr/Readarr/Prowlarr: `/api/v1/…`.
+- Prowlarr is port **9696** — 8090 is Jackett.
+- **Empty body = endpoint absent**, not a failure: probe endpoints by status
+  code first; v1 has no `/api/v1/profile` endpoint at all.
+- `PUT /downloadclient` and `PUT /indexer/{id}` take a SINGLE resource object
+  (GET it, patch `settings`, PUT the whole object back). An array body fails
+  with `Path: $ | BytePositionInLine: 1`.
+- API keys live in each container's config:
+  `podman exec <svc> grep -o "<ApiKey>[^<]*" /config/config.xml | cut -d">" -f2`.
+  They are local-only secrets — they stop being secret the moment a vhost
+  serves the API unauthenticated.
+
+## Schema-first probing (validators are free documentation)
+
+Send the FULL schema template object with fields filled; v1 validators reject
+partials, and each error names the next required field — peel validators one
+live call at a time instead of guessing payload shapes:
+
+- `'Fields' must not be empty` → send a full schema copy, not `fields:[]`.
+- `'Base Url' must not be empty` when keys were camelCase → **PascalCase is
+  required** (`BaseUrl`, `ApiKey`) — schema keys, not the JSON wire keys.
+- `DefaultMetadataProfileId must be > 0` → fetch real ids from
+  `GET /api/v1/metadataprofile` (rows like `(1,'Standard')`);
+  `GET /api/v1/qualitydefinition` returns the quality CATALOG — ids ≠
+  profiles.
+- `targetType null` → configContract not registered in that build (this
+  Prowlarr build has no Prowlarr-plugin indexer; integration is
+  Torznab-family only).
+- `testingPackageId` in a PUT body downgrades validation failures to
+  warnings.
+- `syncLevel` is enum `ApplicationSyncLevel` — omit it or use the int; a
+  string is rejected.
+
+**Verification rules:** `POST /downloadclient` returns the COLLECTION, not
+the created object — verify via GET. Test endpoints return verdicts in the
+BODY at HTTP 200 (`{}` = pass, a list = fail) — judge from bodies, not
+status codes. Command POSTs (`MoviesSearch`, `ApplicationSync`) run
+server-side and respond only when the work finishes: curl `-m` timeouts
+(HTTP 000) during a search are NOT failures — poll queue/history afterward
+instead of re-firing; warm small commands (`RefreshMonitoredDownloads`)
+answer in ~20 ms, so a hang means live work in progress, not a dead API.
+
+**DB as arbiter:** when endpoint shapes are opaque, stop the container and
+edit `/opt/prowlarr/config/prowlarr.db` with `sudo python3` sqlite directly,
+then start. The indexers table columns are
+Id/Name/Implementation/Settings/ConfigContract/Enable/Priority/Added — the
+enable flag column is `Enable` (not `EnableRss`). Prowlarr's real application
+settings values live in `prowlarr.db` → `Applications.Settings` JSON (the
+`/api/v1/applications` GET responses carry `fields: []`).
+
+## Auth posture probe (ground truth is behavior, not config flags)
+
+`AuthenticationMethod=None` + `AuthenticationRequired=Enabled` in
+`config.xml` is ambiguous on paper; on the wire it means a login screen with
+**zero users** — first-account-wins.
+
+1. **Confirm which host runs the service before probing** (`podman ps` on
+   the target, not the box you ssh'd into first — the arrs live on
+   FrankBook-Pro).
+2. Probe an API endpoint that bypasses the SPA shell, locally and through
+   any vhost: `curl -s -o /dev/null -w "%{http_code}"
+   http://<host>:<port>/api/v3/system/status`. **SPA roots (`/`) always
+   serve a 200 shell — never conclude "open" or "gated" from `/`.**
+   401 = gated; 200 + JSON = API fully open (anyone on the net can drive it
+   with curl).
+3. For config truth, dump the whole file
+   (`podman exec <svc> cat /config/config.xml`), not a piped
+   `grep | sed -n 2p`: **`sed -n Np` exits 0 on zero matches**, so the
+   pipeline hides a miss and you conclude "empty field" when the grep itself
+   failed.
+4. Decision framing: when posture = gated-with-zero-users, present it as a
+   decision menu (None behind the front-door wall vs External/authentikarr),
+   not a silent change — first-account-wins means whoever logs in first owns
+   the instance.
+
+## The `None` + `Enabled` auth-config trap (v1 .NET apps)
+
+`<AuthenticationMethod>None</AuthenticationMethod>` +
+`<AuthenticationRequired>Enabled</AuthenticationRequired>` is
+self-contradictory: the UI Save is permanently disabled, and any save throws
+`DryIoc.ScopedItemException` in `AuthorizationMiddleware` → 500 on every
+page after restart. The valid no-auth representation is **absence of the
+`<AuthenticationRequired>` key**, not the literal string `Disabled`. Fix =
+delete the key from `/config/config.xml`
+(`sed -i "/<AuthenticationRequired>/d"`), restart the container. Tell-tale:
+one v1 app boots clean while siblings 500-sprawl → diff their config.xml
+auth blocks; the healthy one shows the correct shape. (Prowlarr runs
+`Forms` + `Enabled` healthily — the crash combo is `None`+`Enabled` only.)
+
+General rules that saved the session these came from: container
+`State=running` ≠ app serving — read `podman logs --tail 12 <c>` before
+theorizing; when a UI gate blocks saving, go under it (these apps keep
+plaintext XML config on disk — read/edit via `podman exec`).
+
+## See Also
+
+- [[Prowlarr]] — indexer feeds and the v2 sync model
+- [[Sonarr]] · [[Radarr]] · [[Lidarr]] · [[Readarr]] — the managers
+- [[qBittorrent]] — download client, 4.6.7 pin, E2E verification
+- [[FrankBook-Pro]] — the host and its podman machine
+- [[Homepage Status Probes]] — how the e2e gate treats this sleeping host

serverAppConfigs-docs/FrankBook-Pro/qbittorrent.md

@@ -111,6 +111,90 @@ Restart=always
   probe one ssh-context out returns the residential IP — a false leak alarm.
 - Ensure umask is set to `022` so downloaded files are readable by arr apps.
 
+## E2E verification after any pipeline change
+
+Run after ANY change to the torrent pipeline — WireGuard credential rotation,
+qB version change, gluetun reconfig, category re-seeding. Proves the full
+grab path (tunnel egress → qB download → category save path → clean
+removal). All commands from a dev box via `ssh frankbook-pro`; inside the
+Mac shell export `PATH="/usr/local/bin:/opt/homebrew/bin:$PATH"` for
+`podman machine ssh`.
+
+**Access note (2026-09-21):** the Mac's login user is **dank**@192.168.0.112
+(NOT core — `core` is the podman-machine VM user, reached only via
+`podman machine ssh`). blankpad has key auth for `dank@` (bootstrapped
+2026-09-21) and a `frankbook-pro` alias in `~/.ssh/config`. macOS sshd has a
+low MaxAuthTries: when password-bootstrapping use `-o
+PreferredAuthentications=password -o PubkeyAuthentication=no
+-o IdentitiesOnly=yes` or the key offers exhaust the attempts before the
+password is tried.
+
+**1. Egress gate — MUST run inside gluetun:**
+
+```bash
+podman machine ssh 'podman exec gluetun wget -qO- https://ipinfo.io/json'
+```
+
+Require `org` = AS9009 (M247 — ProtonVPN Panama PA-12 config); judge by org,
+never country alone. NEVER run this one ssh-context out (bare `wget` on the
+Mac host): it silently returns the residential Comcast IP and reads as a
+hard-gate leak that does not exist. A residential reading ⇒ re-run inside
+the container before diagnosing.
+
+**2. Login — generic cookie parse.** The QPW fetch is flaky (~1 in N returns
+empty); assert length before use — an empty-password run once produced a
+Forbidden loop that cost hours:
+
+```bash
+QPW=$(podman machine ssh 'cat /home/core/Containers/qbittorrent/env' | grep QBIT_WEBUI_PASSWORD | cut -d= -f2)
+[ ${#QPW} -eq 20 ] || exit 1
+SID=$(curl -si --data-urlencode "username=admin" --data-urlencode "password=$QPW" \
+  http://127.0.0.1:8080/api/v2/auth/login \
+  | grep -oE '[A-Za-z_]*SID[A-Za-z0-9_]*=[^;[:space:]]*' | head -1)
+curl -b "$SID" …   # pass the WHOLE match
+```
+
+WebUI from the LAN: `http://192.168.0.112:8080` (published on the gluetun
+unit). Username `admin`; password = the machine-local `QBIT_WEBUI_PASSWORD`
+(20 chars, generated once at provisioning by the deploy handler — never in
+the repo). Parse-bug 403s never reach qB as failed logins — no ban accrues;
+server-side truth is `/Volumes/COMPLETE/qbittorrent-data/qBittorrent/config/qBittorrent/logs/qbittorrent.log`
+(`WebAPI login success/failure` + source IP). Repeated REAL failures accrue
+toward qB's 5-failures/3600 s IP ban.
+
+**3. Add a legal test torrent.** Blender open-movie torrents (~300 MB, e.g.
+Big Buck Bunny) via `POST /torrents/add` with the raw `.torrent` file
+(`-F "file=@…"`) and `category=Finished Torrents`. Never leave a real-looking
+grab behind.
+
+**4. Poll and verify.** Poll `GET /torrents/info` (~20 s): state
+`downloading` → `moving` (INCOMPLETE→COMPLETE move) → `stalledUP`/`uploading`
+at progress=1. `save_path` must be `/Volumes/COMPLETE/Finished Torrents`
+(qB 4.x auto-creates unknown categories at DefaultSavePath; 5.x did NOT
+inherit category paths — one more reason for the 4.6.7 pin). Confirm on
+disk: `ls "/Volumes/COMPLETE/Finished Torrents/"`. Throughput ~17 MB/s is
+normal for the tunnel.
+
+**5. Cleanup — delete only your artifact:**
+
+```bash
+curl -s -o /dev/null -w "%{http_code}" -b "$SID" \
+  --data-urlencode "hashes=<literal-hash>" \
+  --data-urlencode "deleteFiles=true" \
+  http://127.0.0.1:8080/api/v2/torrents/delete   # expect 200
+```
+
+Target the known hash literally — the user's real grabs share the directory
+and MUST be untouched. Verify: hash absent from `/torrents/info`, your dir
+gone, all other entries intact.
+
+**Version-coupling constraints:** qB is pinned to 4.6.7 (Readarr's EOL
+client rejects qB ≥ 5's `204` login at runtime — see the pin rationale
+above). Do not unpin while Readarr exists. After ANY qB version flip,
+regenerate the `WebUI\Password_PBKDF2` line fresh (same password, 100k
+iterations) — a 5.x-written hash line makes 4.6.7 reject every login.
+Rewrite while qB is stopped.
+
 ## See Also
 
 - [[ARR-Stack]] — full *arr stack documentation

serverAppConfigs-docs/Infrastructure/Anubis.md

@@ -131,6 +131,55 @@ ungated), and gitea (git smart-HTTP clients cannot solve JS challenges —
    distinguish challenge from deny at this release; grep the body for
    `Access Denied` / `Making sure you` instead.
 
+## Guard-chain layer diagnosis (a user reports a guarded page "broken")
+
+Use when a guarded page (e.g. the `www.franksdank.cloud` homepage) reads as
+"broken / links gone / not loading" and the fault must be localized to
+server-side or client-side/session. Chain topology:
+`Browser → Caddy TLS edge → Anubis :8923 → Caddy :8016 → Authentik outpost
+→ Caddy :8017 (@home/@data handlers) → app/static`.
+
+Layer-by-layer checks (run from the host via ssh):
+
+1. **Caddy unit + guard containers**: `systemctl --user is-active
+   caddy.service`; `podman ps --format '{{.Names}}\t{{.Status}}' | grep -iE
+   'anubis|authentik|caddy'` — expect `Up … (healthy)`.
+2. **Anubis challenge page validity**: `curl -sk
+   https://www.franksdank.cloud/ --resolve www.franksdank.cloud:443:192.168.0.143`
+   — a healthy challenge contains `"anubis_challenge"` JSON with
+   `difficulty` and the main JS.
+3. **Outpost errors**: `podman logs --tail 40 authentik-outpost 2>&1 | grep
+   -iE 'error|exception|traceback|500|provider'` — routine dramatiq task
+   logs are normal.
+4. **Post-auth handlers from INSIDE the caddy container** (8016/8017 live in
+   its netns — `curl http://127.0.0.1:8017` from the HOST returns `000`
+   connection refused, which is EXPECTED, not a fault):
+   `podman exec caddy sh -c 'wget -q -S -T5 --header "Host:
+   www.franksdank.cloud" -O- http://127.0.0.1:8017/'` and the same for
+   `/services.json`, `/status.json`. Expect 200 + body. (BusyBox wget in
+   this container has NO `-H` flag — use `--header "Host: …"`.)
+5. **Files on disk**: `ls -la /srv/homepage/` and validate `status.json`
+   parses (`python3 -c "import json;json.load(open('/srv/homepage/status.json'))"`).
+
+Tool/behavior notes: the anubis container is distroless — `podman exec anubis
+sh` fails (`sh: executable not found`); inspect via its logs or the served
+challenge page. Anubis logs `x-real-ip: 10.89.1.22` (the Caddy container's
+podman IP) for ALL clients — rootless-podman published-port SNAT;
+clients are indistinguishable by IP and clearance is cookie-based. Do NOT
+mistake that for the bug. Homepage render: `app.js` does
+`Promise.all([fetch services.json, fetch status.json])`; on ANY fetch
+failure the catch sets `#summary` to `status unavailable: <err>` and the
+grid stays EMPTY — which a user reads as "links gone / not loading." A
+30-service status.json with a 9-service services.json renders fine
+(stateByName lookup).
+
+**Decision:** if every layer is healthy (challenge valid, outpost clean,
+handlers 200, files render), the fault is client-side/session. STOP
+thrashing the live system — repeated failed browser challenges share the
+masked 10.89.1.22 identity. Get the authenticated view: ask for the exact
+symptom (challenge spinning? login loop? empty grid? "status unavailable"
+text?) or the authentik login to drive the page in a real browser.
+
 ## Verify live
 
 ```

serverAppConfigs-docs/Infrastructure/Authentik.md

@@ -342,6 +342,168 @@ The silent re-auth path was proven end-to-end: after deleting
 `qa-e2e-admin`'s rows, the e2e admin spec ran against the stale storage
 state and passed with no login prompt (2 tests, 2.9 s).
 
+## Guard-chain flip recipe (adding a guarded vhost)
+
+Move a public vhost from `reverse_proxy <app>:<port>` onto the chain:
+client → TLS edge → Anubis :8923 → outpost decision → internal
+`http://:8017` listener → Host-matched handler → real upstream.
+
+**Order (never reorder):**
+
+1. **Provider first.** Create the outpost ProxyProvider before touching
+   routes, or authed traffic hits the outpost error page instead of the app.
+   Fleet convention for `external_host` is the exact browser-facing origin
+   (the inventory table carries both bare origins and trailing slashes;
+   routing matches on hostname, verified 2026-09-02).
+   `internal_host = http://caddy:8017`, all three flows set, and an
+   Application attached — the two provisioning rules above.
+2. **Route flip in ONE commit/deploy**: the vhost becomes
+   `reverse_proxy anubis:8923` with `header_up X-Real-Ip {remote_host}`, and
+   inside the `http://:8017` listener add `@<svc> host <vhost>` +
+   `handle @<svc> { reverse_proxy <upstream> }`. Anubis guard set, Caddy TLS
+   edge, `:8017` routing table, and provider list change together.
+3. Deploy via the repo deploy flow (commit first; deploy pushes main itself).
+
+**Verification (all three):**
+
+- Unauthenticated probe from a client: `curl -s https://<vhost>/` → HTTP 200
+  challenge page; grep the body for `challenge|not a bot|anubis` (~8
+  markers on a live challenge). A 502 here instead of the challenge = chain
+  mis-wired.
+- Post-auth routing proof: grep the LIVE config inside the container —
+  `podman exec caddy sh -c "grep -n -A3 <handler> /etc/caddy/Caddyfile"` —
+  after the deploy prints its reload line; confirm the reload actually ran,
+  don't trust exit 0.
+- Outpost reachability: `podman exec authentik-server getent hosts caddy`
+  resolves the compose DNS alias (10.89.1.x) that `internal_host` dials.
+
+**Pitfalls:** probing `http://localhost:8017/` from the host returns HTTP
+000 (connection refused) — the post-auth listener lives in the caddy
+container's netns; probe inside it or grep the config. Non-browser clients
+cannot solve the JS challenge — before flipping, audit service-to-service
+traffic: ingestion/push paths must target internal addresses (`loki:3100`),
+never a public hostname of a guarded service. Exempt-by-construction
+categories stay unguarded: the IdP itself (`auth.`), git smart-HTTP
+endpoints, services with their own complete auth — document the exemption
+in the vhost comment.
+
+## Driving the ORM (`ak shell`)
+
+All authentik object work happens through the server container's ORM shell —
+no repo change, no deploy (these providers are NOT repo-managed; capture new
+provider+application pairs as a blueprint or setup script to avoid drift).
+
+```bash
+ssh dankclaw 'podman exec -i authentik-server ak shell' <<'EOF' | sed -n '/229 objects/,$p'
+from authentik.core.models import User, Group
+...
+EOF
+```
+
+- The heredoc IS the program: `ak shell` reads all of stdin and execs it
+  (heredoc-through-ssh quoting breaks; the raw-stdin pattern is the proven
+  one).
+- Bootstrap prints JSON log lines, then
+  `### authentik shell (2026.8.3)` and `229 objects imported automatically…`
+  — pipe through `sed -n '/229 objects/,$p'` to drop the noise.
+- 229 models are pre-imported. Import paths (each wrong guess costs a failed
+  exec): `User`, `Group`, `Application`, `ApplicationEntitlement`,
+  `PropertyMapping` → `authentik.core.models`; `ProxyProvider` →
+  `authentik.providers.proxy.models` (NOT `authentik.outposts.models`);
+  `LDAPProvider` → `authentik.providers.ldap.models`;
+  `authentik.core.models.application` does NOT exist as an import path; there
+  is no `authentik.policies.group` module in this build.
+- For live DB truth prefer the postgres container
+  (`podman exec authentik-postgres psql -U authentik -d authentik -At -c
+  "select * from …"`) — **query with `select *` and stop guessing column
+  names**: 2026.x renamed columns (`Provider.objects.filter(type=…)` raises
+  FieldError — `type` moved to subclass tables; the outpost `enabled` column
+  may not exist under that name; provider pk is `oauth2provider_ptr_id`).
+
+## User provisioning playbook (user-level account)
+
+There is no `User.create_user` classmethod, and `is_superuser` is NOT a
+constructor kwarg (it is a property derived from group membership) — passing
+it raises `TypeError: User() got unexpected keyword arguments`:
+
+```python
+from authentik.core.models import User, Group
+u = User(username="ridley", email="ridley@franksdank.cloud", name="Ridley", is_active=True)
+u.set_password("...")        # password arrives on stdin only, never argv
+u.save()
+for g in ("Non-Admin Users", "jellyfin_users", "openwebui_users"):
+    u.groups.add(Group.objects.get(name=g))
+```
+
+`path` defaults to `users` (regular). The access map (verified live
+2026-10-04): there are NO per-app policy bindings on the proxy providers
+(all use the implicit-consent "Authorize Application" flow), so guarded-vhost
+access is decided entirely by GROUP MEMBERSHIP consumed downstream —
+
+- `homepage-admins` → Caddy RBAC for homepage `/admin` (`X-Authentik-Groups`
+  regex gate) and the homepage admin panel; regular users get only
+  `category: user` panels from `services.json`.
+- `openwebui_users` / `openwebui_admins` → ApplicationEntitlements
+  "Open WebUI Users"/"Open WebUI Admins" (bound by group), consumed by the
+  "Open WebUI roles" property mapping → `roles` claim. Verify live with
+  `u.app_entitlements(app)`.
+- `jellyfin_users` / `jellyfin_admins` → Jellyfin via the LDAP outpost
+  (filter `(memberOf=cn=jellyfin_users,ou=groups,dc=franksdank,dc=cloud)`).
+- `Non-Admin Users` = the plain-user marker group (precedents: alex,
+  qa-e2e).
+- EXCLUDE from a user-level account: `homepage-admins`, `jellyfin_admins`,
+  `openwebui_admins`, `authentik Admins`, `authentik Read-only`.
+- Other user-category vhosts (SearXNG, Gitea, Calibre-Web, RomM, Copyparty,
+  Plex, Seerr, Marreta) need no per-user group: any authenticated user
+  passes the outpost.
+
+Verify login behaviorally — `authenticate()` is the exact PasswordStage code
+path:
+
+```python
+from django.contrib.auth import authenticate
+assert authenticate(request=None, username="ridley", password="...")   # -> User
+assert authenticate(request=None, username="ridley", password="nope") is None
+```
+
+Look flows up by slug, not name:
+`Flow.objects.get(slug="default-authentication-flow")` (its `.name` is
+"Welcome to authentik!" — `.get(name=…)` fails). Stage order:
+Identification(10) → Password(20) → AuthenticatorValidate(30; no-op with no
+enrolled device) → UserLogin(100). With no MFA enrolled, nothing else blocks.
+
+More field gotchas (each cost a failed exec): `ApplicationEntitlement.application`
+can be null (guard with `getattr`); `LDAPProvider` has
+`search_mode`/`bind_mode`/`base_dn`, lacks `search_base`/`of_groups`/
+`organization`; `OAuth2Provider` has no `.scopes`; `Outpost.config` is an
+`OutpostConfig` object, not a dict; `Policy` has no uniform `.policy_cls`
+(use `type(p).__name__`); `User` has no `.status`.
+
+Never rotate `AUTHENTIK_SECRET_KEY` to fix claim issues — it invalidates
+every outpost session fleet-wide; purge the stale `proxysession` rows
+instead (section above).
+
+## Probing the outpost from inside the netns
+
+Rootless podman gives each container its own netns; **host `ss -ltn` cannot
+see listeners inside other containers' namespaces** — a missing listener on
+the host is an observation artifact, not a fault. Distroless images have no
+shell (`podman exec authentik-server sh` fails), so probe by joining the
+namespace:
+
+```
+podman run --rm --network container:<container> --entrypoint /bin/busybox \
+  docker.io/library/busybox:latest \
+  wget -q -T 4 -t 1 -O /dev/null -S "<url>" 2>&1 | head -8
+```
+
+- Quote URLs over ssh: the remote shell is zsh and bare `?` in query strings
+  triggers globbing (`no matches found`).
+- Boot transience: immediately after a restart the same probe can return 503
+  (configure task pending); poll until the answer changes before concluding.
+- `podman inspect --format "{{.Config.NetMode}}"` errors on this podman
+  version — the field is `.HostConfig.NetworkMode`.
+
 ## Failure modes
 
 | Symptom | Cause | Fix |

serverAppConfigs-docs/Infrastructure/Homepage Status Probes.md

@@ -0,0 +1,158 @@
+---
+type: reference
+title: Homepage Status Probes
+description: The homepage TUI status page — how probe targets are chosen (host-netns vantage), the three false-down shapes, triage procedure, local render check, deploy, and the e2e gate interactions (asleep Mac mini, raw-IP matrix).
+tags: [homepage, status, probes, e2e, homepage-status]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Homepage Status Probes
+
+The homepage is a static site —
+`shared/homepage/{index.html,admin.html,app.js,style.css,services.json,docs.html}`
+— served from `/srv/homepage` on **dankclaw** only (`deploy_homepage` in
+`scripts/deploy-remote.sh` refuses other targets). A checker
+(`scripts/homepage-status.py`, run by the `homepage-status.timer` unit) probes
+each card's target and writes `/srv/homepage/status.json` (mode **0644** so
+Caddy can serve it — losing the mode 403s the page).
+
+## The checker's vantage and its three false-down shapes
+
+The checker probes TCP/HTTP **from the host netns**. It only has three states
+— `wip` | `up` | `down`; there is no "unknown", so an unprobeable service can
+never render honestly and must be dropped or re-pointed, not "fixed".
+
+1. **TLS-only app probed with `method:"http"`.** Plex Media Server serves TLS
+   only on `32400`; plain-HTTP GET → TLS alert → curl exit 52, code `000` →
+   card says down while PMS answers `200` over HTTPS. Fix:
+   `"method":"tcp"` (connect is the honest liveness signal; the public URL
+   stays HTTP because Caddy re-keys).
+2. **Retired port.** The Hermes card probed `127.0.0.1:8090` after the
+   compose→quadlet move moved the dashboard to `9119` and the webui to
+   `8787`; nothing listened on 8090, so a healthy stack read `down`.
+3. **Unpublished port / wrong netns.** `gitea-postgres` (`db:5432`, net
+   `systemd-gitea`), `searxng-valkey` (`valkey:6379`, net `systemd-searxng`)
+   and `alloy` (OTLP 12345) publish **no** ports. Container DNS aliases are
+   netns-scoped, so from the host these are unreachable *by construction* —
+   no probe value makes them honest. Drop those cards.
+
+A host-vantage probe can only reach a listener **published to the host**
+(rootlessport/pasta); unpublished ports are invisible from the host netns.
+See [[Container Host Reachability]] for the netns rules.
+
+## Triage procedure
+
+1. Read the card's probe target in `shared/homepage/services.json`; note
+   method (`tcp` vs `http`).
+2. Does that port have a **host-visible listener**?
+   `podman ps -a --format '{{.Names}}|{{.Status}}'`, then per container
+   `podman inspect <c> --format '{{json .NetworkSettings.Networks}}
+   {{.HostConfig.NetworkMode}}'` and `podman port <c>` (empty output =
+   published nothing → structurally unprobeable from the host).
+3. Probe from the **correct vantage**: from a bridge-net container the host
+   is its gateway IP (`/proc/net/route`, e.g. `10.89.1.1` on
+   `dankclaw_default`) — *not* `host.containers.internal` (`169.254.1.2`),
+   which belongs to the default `podman` net (`10.88.0.0/16`). Probe with
+   bash `/dev/tcp` (no `python3`/`ip` in these images); escape loop vars so
+   they expand **inside** `podman exec bash -c`.
+4. Classify: real outage (container exited — read `State.ExitCode`: 137 =
+   SIGKILL, 143 = SIGTERM/manual stop; `OOMKilled=false` rules out the OOM
+   killer) vs false down (running + unreachable target).
+5. Fix: re-point or re-method for HTTP/TLS mismatches; **drop** cards whose
+   only exposure is an internal netns; never invent a probe target.
+6. Verify by re-running the checker path, not by eyeballing: after deploy,
+   re-read `/srv/homepage/status.json` and diff the `down` list
+   before/after.
+
+Removing a card's `probe` key without a code change crashes the checker
+(`svc["probe"]` KeyError) — dropping the whole entry is the only zero-code
+fix. The `description` field is consumed only by `app.js` `descHTML()`, so
+adding a `details` object (role/spec/inputs/use) is schema-safe:
+`tests/homepage_config.bats` checks only name/category/section/probe.
+
+## Local render check (before every deploy)
+
+The page fetches `services.json` + `status.json` at load; `file://` is
+CORS-blocked, so serve over HTTP:
+
+```bash
+cd shared/homepage
+python3 -c "import json; s=json.load(open('services.json')); \
+  json.dump({'services':[{'name':x['name'],'state':'up'} for x in s]}, open('status.json','w'), indent=2)"
+python3 -m http.server 8899 --bind 127.0.0.1
+```
+
+Then in the browser: load `http://127.0.0.1:8899/index.html`, **wait ~1.5 s**
+(panels render after the async fetch; checking immediately returns 0 panels),
+extract via `tab.evaluate` — LLM cards render `.desc-role`/`.desc-spec`/
+`.desc-inputs`/`.desc-use` from the card's `details` object, other cards a
+plain `.desc`; `getComputedStyle(el).color` confirms tones. Screenshot for
+visual proof. Expected local noise: `deployed.js` 404s (baked in at deploy
+time) — harmless.
+
+The scratch `status.json` is untracked and NOT git-ignored — move it to
+`shared/homepage/.AGENTTRASH/` after the check (repo policy: never `rm`).
+
+CSS palette lives in `style.css` `:root` (`--fg` bright green, `--dim` dim,
+`--head` pale, `--link` cyan, `--wip` yellow, `--down` red). Add new tones as
+`--*` vars and include any new `.desc-*` class in the `html[data-bare="1"]`
+reset so `?bare=1` strips it to black.
+
+## Deploy
+
+```
+./scripts/deploy.sh dankclaw --service homepage --skip-e2e
+```
+
+Copies the assets to `/srv/homepage`, regenerates `deployed.js` (git commit
+datetime), restarts the `homepage-status` timer. Verify live:
+`ssh dank@192.168.0.143 'grep … /srv/homepage/…'`.
+
+**`--skip-e2e` is justified for homepage-only changes when the Mac mini is
+asleep**: the e2e admin crawl (`tests/e2e/specs/11-admin.spec.ts:24`) walks
+Prowlarr at `192.168.0.112:8989` on FrankBook-Pro; an asleep mini (ARP STALE,
+no remote wake) times out that test — a pre-existing environmental failure,
+not a regression. All non-Mac-mini e2e tests pass.
+
+## e2e raw-IP *arr failures — probe from the browser's box, not a peer
+
+When `11-admin` fails with `ERR_CONNECTION_TIMED_OUT`, the instinct is "the
+Mac is asleep". Sometimes true; not always. Playwright runs on the dev box,
+so a healthy probe from dankclaw proves nothing about the **browser's** path:
+
+```bash
+# 1. From the box that RUNS the browser:
+curl -s -o /dev/null -w '%{http_code}\n' --max-time 7 http://192.168.0.112:8989/
+# 2. From a LAN peer:
+ssh dankclaw 'for p in 8989 7878 9696; do curl -s -o /dev/null -w "$p %{http_code}\n" \
+  --max-time 6 http://192.168.0.112:$p/; done'
+# 3. Liveness (ICMP, not TCP):
+ping -c 2 -W 3 192.168.0.112
+```
+
+| peer probe | dev-box probe | ping | meaning |
+|---|---|---|---|
+| 200 | 200 | ok | transient — re-run the spec |
+| timeout | timeout | fails | Mac asleep; wake, ports return in seconds |
+| **200** | **rc=28** | **ok** | **path problem, not sleep** — ICMP allowed, TCP blocked. Waking cannot help |
+
+Third row: confirm the Mac is innocent through the dankclaw jump
+(`pgrep -fl socat`; `netstat -an | grep -c 8989`; loopback
+`curl http://127.0.0.1:8989/` → 200; `/usr/libexec/ApplicationFirewall/socketfilterfw
+--getglobalstate` → disabled). Forwarders up + loopback 200 ⇒ the block is
+between the dev box and the Mac.
+
+Don't hold a deploy hostage to a known-environmental red gate: verify the
+deployed service **directly** (its own ledger/health/real completion), state
+the gate was skipped or partially red and why (`--skip-e2e` prints the
+warning, so the skip is recorded), and report the residual gap plainly
+("29/30 e2e pass; 11-admin fails on Sonarr's raw IP from the dev box; peer
+probe returns 200; deploy verified by other means"). Do not weaken or skip
+the spec — it guards a real SSO-plus-RBAC contract.
+
+## Related
+
+- [[Container Host Reachability]] — the netns rules behind the vantage
+- [[Observability Provisioning]] — the monitoring stack reading the same fleet
+- [[Deploy State Triage]] — service-scoped deploys and the e2e gate
+- [[Testing Policy]] — the seam/e2e contract this page's checks live in

serverAppConfigs-docs/Infrastructure/LAN-DNS.md

@@ -158,6 +158,119 @@ The Porkbun zone `franksdank.cloud` carries an A record for every vhost plus the
 | HTTPS works from the host but not the WAN | No port forward on the G36 for that service (manual router step) |
 | Cert renewal fails in Caddy logs | Porkbun API credentials in the Dankclaw `.env`; test with the Porkbun API `/ping` endpoint |
 
+## Per-client fix: resolver drop-in + browser DoH bypass
+
+Symptom: a LAN machine gets `connection timeout` (curl rc=28) on
+`https://*.franksdank.cloud` while the server is healthy. Root cause: the
+public DNS hands out the WAN IP `71.227.227.172`, which has **no 443
+port-forward** (filtered). Any resolver that is NOT Pi-hole produces the WAN
+answer → SYN to a filtered port → timeout. There are TWO resolver layers to
+check, in order:
+
+**Layer 1 — systemd-resolved (fixes curl/shell).** A route-only drop-in
+sends `*.franksdank.cloud` ONLY to Pi-hole and never to the default-route
+servers:
+
+```bash
+dig +short www.franksdank.cloud @192.168.0.143   # expect 192.168.0.143
+dig +short www.franksdank.cloud @1.1.1.1         # expect 71.227.227.172 (the trap)
+getent hosts www.franksdank.cloud                # what the OS actually resolves
+curl -sk --resolve www.franksdank.cloud:443:192.168.0.143 \
+     https://www.franksdank.cloud/ -o /dev/null -w '%{http_code}\n'   # 200 = server fine
+
+sudo mkdir -p /etc/systemd/resolved.conf.d
+sudo tee /etc/systemd/resolved.conf.d/franksdank-lan.conf >/dev/null <<'EOF'
+[Resolve]
+DNS=192.168.0.143
+Domains=~franksdank.cloud ~franksdank.casa
+EOF
+sudo systemctl restart systemd-resolved && resolvectl flush-caches
+```
+
+The `~` (route-only) prefix is load-bearing: it routes only those domains to
+Pi-hole. Verify: `getent hosts www.franksdank.cloud` → `192.168.0.143`;
+`resolvectl status` shows `DNS Domain: ~franksdank.cloud ~franksdank.casa`.
+
+**Layer 2 — browser DoH (the bypass: "curl works, browser times out").** If
+Layer 1 is verified but the BROWSER still times out, the browser is doing
+its own DNS over HTTPS and never asks systemd-resolved — it queries the TRR
+provider (Cloudflare et al.) directly → WAN answer → timeout. Prove it:
+`curl -sk -m 6 https://71.227.227.172/` → rc=28 (the WAN answer is a dead
+end). Diagnosis details:
+
+- Firefox profile may be at the XDG path `~/.config/mozilla/firefox/`, NOT
+  `~/.mozilla/firefox/` — find it via the running process if the default
+  path is empty.
+- `grep network.trr. <profile>/prefs.js` — **no entries = shipped default,
+  and since Firefox 113 the shipped default is DoH ON.**
+
+Fix (surgical, keeps DoH for everything else): Firefox → Settings → Privacy
+& Security → **DNS over HTTPS** → "enabled for these domains" → **Add…** →
+`franksdank.cloud` (and `franksdank.casa`). The exception list = domains
+that do NOT use DoH; they fall back to the system resolver, which has the
+route-only drop-in. Or set DoH to **Off** entirely (coarser). Chrome:
+Settings → Privacy and security → Security → "Use secure DNS" → off, or
+"Use the same settings as your system". Do NOT hand-edit `prefs.js` while
+the browser is running — it holds prefs in memory and rewrites the file on
+exit, clobbering the edit.
+
+This is a per-client fix: every LAN client that must reach the homelab by
+name needs Pi-hole as its DNS (or the drop-in) AND a browser that does not
+bypass it with DoH. The server side needs no change. Verified on blankpad
+2026-09-21: the drop-in was present since 09-12, curl returned 200, and
+Firefox 155 with default (ON) DoH was the remaining hole.
+
+## Porkbun zone ops (franksdank.cloud)
+
+Operate the public DNS zone via the Porkbun API v3, run **on Dankclaw** so
+the API key/secret sourced from `/home/dank/serverAppConfigs/.env`
+(`PORKBUN_API_KEY`, `PORKBUN_API_SECRET_KEY`) never leaves the host. Pipe
+scripts to the target with `ssh -o BatchMode=yes dankclaw 'bash -s' <
+/tmp/script.sh`.
+
+Endpoints (all POST, JSON body):
+
+- List: `https://api.porkbun.com/api/json/v3/dns/retrieve/franksdank.cloud`
+  — auth via headers `X-API-Key` + `X-Secret-API-Key` (Bearer is wrong
+  here). Response shape is **top-level `.records[]`**, NOT
+  `.content.records[]` as some docs claim — parse with
+  `jq -r '.records[] | [.id,.name,.type,.content,(.ttl|tostring)] | @tsv'`.
+- Delete: `/dns/delete/{domain}/{id}` (body carries apikey+secretapikey).
+- Edit: `/dns/edit/{domain}/{id}` — pass the full record (`id`, `name`,
+  `type`, `content`, `ttl`, `prio`).
+
+Quirks that bite (verified live 2026-08-31):
+
+1. **`name` is a RELATIVE label on edit.** Passing `ai.franksdank.cloud`
+   yields the record `ai.franksdank.cloud.franksdank.cloud`. Pass the bare
+   label (`ai`) and the API appends the zone itself. Verify after every
+   rename with a retrieve before declaring done.
+2. **Rate limit answers as an HTML 404-style page, not JSON.** A batch of
+   rapid mutations makes some calls return `<!DOCTYPE html>...` where jq
+   then errors (`Cannot iterate over null` / `Invalid numeric literal`). Do
+   not assume the call succeeded — sleep ~8 s and retry the failed ids;
+   verify by re-listing the zone at the end.
+3. Deletes/renames are per-record-id: always retrieve first, map name→id,
+   then mutate by id.
+
+Rename procedure (vhost rename = DNS + Caddy together):
+
+1. `retrieve` → collect ids for the renamed names.
+2. `edit` each id with the new **relative** label, preserving
+   type/content/ttl.
+3. Re-retrieve and confirm `<new>.franksdank.cloud A <ip>`.
+4. In-repo lockstep before deploy: rename the Caddyfile vhost block AND the
+   Pi-hole `dankclaw/pihole/addn-hosts` line in the same commit — a test
+   fails the suite if any Caddyfile vhost lacks an addn-hosts A record on
+   192.168.0.143.
+5. Deploy via `scripts/deploy.sh dankclaw` (plans pihole+caddy from changed
+   paths). Capture the suite's true exit with
+   `out=$(./tests/run.sh 2>&1); rc=$?` — piping to `tail` masks the exit
+   code behind `tail`'s own status and has shipped untested commits here.
+6. Verify live: `dig +short <name> @192.168.0.143` then
+   `curl -o /dev/null -w %{http_code} https://<new-name>` (expect 200/302;
+   first hit triggers dns-01 issuance, which works without inbound port 80).
+
 ## See Also
 
 - [[Pi-hole]] — the LAN DNS server (quadlet, records, web UI)

serverAppConfigs-docs/Infrastructure/Observability Provisioning.md

@@ -0,0 +1,95 @@
+---
+type: reference
+title: Observability Provisioning
+description: How Prometheus config/rules and Grafana datasources/dashboards are provisioned as code in serverAppConfigs — the config triad, deploy copiers, planner mappings, seam tests, entry points for each change type, and API verification.
+tags: [observability, prometheus, grafana, provisioning, deployment]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Observability Provisioning
+
+The monitoring stack is **provisioned as code**: dashboards, datasources,
+scrape jobs and alert rules live in the repo and ride the normal deploy flow.
+No hand-edits inside containers — they are wiped on recreate and never
+committed.
+
+## The config triad
+
+| Repo path | What it holds |
+|---|---|
+| `shared/prometheus/prometheus.yml` | scrape jobs (global + per-host) |
+| `shared/prometheus/rules/llm-fleet.yml` | alert rules + recording rules |
+| `shared/grafana/provisioning/` | datasources + dashboards (provisioned tree) |
+
+Targets and datasources reference **container names** (`loki:3100`,
+`alloy:12345`), never loopback-published ports; the LLM endpoints are reached
+by LAN IP (`192.168.0.143:7770`, `:7771`, `192.168.0.235:7770`) because they
+are host-netns services, not containers. Prometheus 3 refuses HTML responses
+as metrics — HTML-scraped "probes" are structurally inert and need a blackbox
+exporter (see [[Stale Signal Audit]]).
+
+## The deploy path
+
+`scripts/deploy-remote.sh` carries two copier functions:
+
+- `copy_prometheus_config` → `/opt/prometheus/config/` on the target
+- `copy_grafana_provisioning` → `/opt/grafana/provisioning/`
+
+`scripts/deploy-plan.py` maps changes under `shared/prometheus/**` and
+`shared/grafana/**` to the `prometheus`/`grafana` services, so
+`deploy.sh dankclaw --service grafana` (etc.) restarts only what the change
+touched. Seam tests in `tests/observability_quadlet.bats` pin: unit files
+exist, the copier functions exist, and the provisioned paths match the units'
+mounts.
+
+## Entry points
+
+**Add a scrape target** — edit `shared/prometheus/prometheus.yml`. Decide the
+vantage first: container name if the exporter is a container on the same
+netns; host LAN IP if it is a host process (LLM endpoints); a blackbox module
+if it is an HTTP probe, not a metrics endpoint.
+
+**Add an alert/recording rule** — edit `shared/prometheus/rules/llm-fleet.yml`.
+Current delivered set: 5 alert rules + 1 recording rule for the LLM fleet.
+Rules reload with the prometheus service.
+
+**Add a datasource** — drop a YAML file into
+`shared/grafana/provisioning/datasources/`. Point it at the container name of
+the backend (`http://prometheus:9090`), not loopback.
+
+**Add a dashboard** — drop the JSON into
+`shared/grafana/provisioning/dashboards/`. Provisioned dashboards live in
+unified storage, not (only) the SQLite table — grep the API, not the DB, when
+verifying (see the verify section of [[Fleet Audit and Posture]]).
+
+## Verify through the APIs, not the filesystem
+
+```bash
+# prometheus: rules and targets actually loaded
+curl -s http://127.0.0.1:9090/api/v1/rules | python3 -m json.tool | head
+curl -s "http://127.0.0.1:9090/api/v1/targets?state=active" | \
+  python3 -c "import json,sys; [print(t['labels'].get('job'), t['health'], (t['lastError'] or '')[:60]) for t in json.load(sys.stdin)['data']['activeTargets']]"
+# grafana: datasource + dashboard provisioned
+curl -s -u admin:<pw> http://127.0.0.1:3002/api/datasources | python3 -m json.tool | grep -E '"name"|"url"'
+```
+
+Health answers from a stale process and provisioned resources can hide in
+unified storage — a file existing on disk is not proof the running server
+loaded it. See [[Stale Signal Audit]] for the never-green-target decision
+table (refused = outage, 404/HTML = config defect).
+
+## Current delivered state (2026-09-17)
+
+Prometheus → Alertmanager (`dankclaw/observability/quadlet/alertmanager.container`,
+loopback `:9093`) → host-side Signal relay (`scripts/signal-relay.py`, user
+unit, `192.168.0.143:18093` — the LAN IP, the one bind a container can reach
+via `host.containers.internal`) → signal-cli. Some targets read down for
+structural reasons (HTML-scraped "probes", Gitea metrics off); the remedy is
+the blackbox exporter. See [[Observability]].
+
+## Related
+
+- [[Observability]] — stack topology and the alert delivery chain
+- [[Stale Signal Audit]] — classifying down targets before "fixing" them
+- [[Quadlet Lifecycle]] — the units that host these services
+- [[Deploy State Triage]] — how the planner selects these services

serverAppConfigs-docs/LLM/DFlash DSpark Speculative Decoding.md

@@ -95,6 +95,51 @@ llama-server --hf-repo <target-GGUF> \
 
 The DFlash drafter proposes, then ngram-mod runs as an additional implementation in the chain. Verified against upstream `common/speculative.cpp` (b10133+).
 
+### Chaining laws (measured on this fleet, 2026-09 → 10)
+
+`--spec-type` takes a comma list, and the gate is `ctx_dft` in
+`common/speculative.cpp` (`common_speculative_init_result`): **draftless**
+types (the ngram family — simple, map-k, map-k4v, mod, cache) are added
+unconditionally, while **context-backed** types (`draft-mtp`, `draft-dflash`,
+`draft-dspark`, `draft-eagle3`) all require a draft CONTEXT. Consequences:
+
+- A chain may hold ONE context-backed drafter plus any number of ngram types.
+  `draft-dflash,ngram-mod` is legal and measured: 2.6x the drafter alone on
+  prompt-echo traffic, byte-identical to it on prose at the fleet's 48/64
+  window.
+- Two context-backed types cannot share one process: `draft-mtp,draft-dflash`
+  dies at load with `context type MTP requested but model doesn't contain MTP
+  layers` — MTP builds its context against whatever the draft context holds,
+  so it met the DRAFTER's file, not a missing head. Pin that in a seam so
+  nobody "fixes the flag" later.
+- Internal order is fixed (ngram types first, then drafters); CLI order is
+  irrelevant.
+- **A legal chain can still lose on a recurrent model.** `draft-mtp,ngram-mod`
+  on the qwen35 hybrid (frank-large era) measured −15 % prose / −7 %
+  block-copy (2026-10-03): adding the ngram layer logs `recurrent state
+  rollback is not compatible with 'ngram-mod' - disabling rollback support`,
+  and MTP verification depends on that rollback, so acceptance drops on ALL
+  traffic even where the matcher fires. Measure chains per architecture,
+  never by analogy.
+- ngram-mod prints NO startup banner — the load log shows only the drafter's
+  `adding speculative implementation` line. Never conclude "chain not
+  engaged" from the load log; identify behaviourally (`draft_n` > 0 only
+  where text repeats) or by the per-task `draft acceptance = ... mean len`
+  line. It costs ~64 MiB of host-RAM hash table.
+
+### Drafter KV: read the window policy from the header
+
+The draft context is built with `cparams.n_ctx = llama_n_ctx(ctx_tgt)`
+(`common/speculative.cpp`) — there is no flag to cap it — so a drafter that
+declares NO sliding window caches the whole target window on whatever device
+`--spec-draft-device` names. DFlash declares `sliding_window = 2048` (~40 MiB
+of draft KV); DSpark declares none: 5 layers x 2 x 8 kv heads x 128 dim x 2 B
+= 20 KiB per position, i.e. **5.0 GiB at 262144** — enough to kill the first
+draft graph capture with `ROCm error: out of memory` AFTER the model loaded
+and `/health` answered (measured 2026-09-16). Read `sliding_window` out of
+the drafter's GGUF header before sizing any spec-decode deployment, and
+prefer a skewed `-ts` over a context cut when the full window matters.
+
 DFlash on llama.cpp:
 
 ```
@@ -209,6 +254,151 @@ docker run --rm --name vllm-qwen36-dflash \
 
 History: before `--spec-default`, the box ran DSpark (`draft-dspark`, retired 2026-08-15 after a ROCm illegal-memory-access crash under two concurrent 16k prompts) and before that bundled MTP (`draft-mtp`, retired same day). DFlash was retired on 2026-08-14 (premature-EOS under parallel load). The old DFlash draft GGUF (`Qwen3.6-27B-DFlash-Q8_0.gguf`), the MTP-only drafter (`Qwen3.6-27B-MTP-ONLY-Q8_0.gguf`), and the DSpark draft (`Qwen3.8-27B-DSpark-Q8_0.gguf`) remain in `/home/dank/models/` as reference only.
 
+## A/B protocol on a live llama.cpp endpoint
+
+Verdicts are **traffic-shaped**: a proposer can only draft text it has seen,
+so a single prompt shape answers the wrong question. Use fixed shapes:
+
+| Shape | Prompt | Who it can favour |
+|---|---|---|
+| prose | 1k filler + "reply with one sentence" | model-backed drafters only |
+| echo | ~1.5k passage of UNIQUE lines + "repeat verbatim" | ngram (output IS the prompt) |
+| structured | repetitive output, no prompt overlap | ngram, if spans really repeat |
+| identity | short deterministic question | controls for lossless verification |
+
+Design the echo passage so no long span repeats INSIDE it — otherwise you
+measure the matcher against itself. For `structured`, give each line a unique
+id: repeated lines make it a negative control, not a repetition test.
+
+Protocol:
+
+1. Stop the unit (`sudo systemctl stop <svc>`), `pkill -x llama-server`; wait
+   for the port (a retired predecessor's published-port proxy leaves the
+   endpoint port in TIME_WAIT — the bind fails with address-in-use long after
+   the unit stopped).
+2. Per config: write an env file with the overrides, start the serve script
+   with `FRANK_*_ENV_FILE`, poll `/health`, measure. Ready-made harness:
+   `scripts/bench-spec-ab.py <env-file> <serve-script>`.
+3. **Restore the unit in a `finally`** and shout if it does not come back
+   healthy. A crashed harness that leaves production down is worse than no
+   harness.
+4. **Prove losslessness**: one fixed prompt at temp 0 to every config, diff
+   the greedy text. Speculation only buys speed; differing texts are a
+   correctness bug. Empty answers (`content: null` when reasoning ate the
+   budget) make the comparison vacuous — bench clients must fall back to
+   `reasoning_content`, and echo/structured prompts should disable thinking
+   per request: `chat_template_kwargs: {"enable_thinking": false}`.
+5. **Sweep the depth, not just the mode** — draft depth is model-specific and
+   single-valued; deeper often loses (measured: at n=7 a third of proposals
+   survived, at n=3 more than half).
+6. **Sweep the ngram window in the same run as the configs**: too short a
+   `--spec-ngram-mod-n-min/-n-max` proposes on ordinary prose fragments —
+   rejected, verification still paid (measured 43.9 → 36.0 tok/s).
+
+Read the right counters — per request `timings.draft_n` / `draft_n_accepted` /
+`predicted_per_second`; cumulative `/metrics`
+`llamacpp:spec_decode_num_drafts_total` / `..._accepted_tokens_total`;
+per-task `draft acceptance = ... (N accepted / M generated)`. `draft_n == 0`
+while tokens flow = the proposer is NOT engaged, whatever the flags say. A
+zero-draft config is a useful negative control: it proves the other config's
+drafts came from the proposer under test.
+
+Harness hygiene: absolute script paths (a relative path plus `cd` to its own
+dir double-counts the prefix); `pkill -x`, never `-f` (the pattern matches
+your own command line); nested f-strings with backslash escapes are a
+SyntaxError on the fleet's Python (<3.12 semantics) — build log lines with
+concatenation. Report `decode tok/s` per shape plus acceptance and mean
+accepted length; a config that wins on echo and loses on prose is a traffic
+decision, not a winner. Wire the winner into the deployed env file with the
+measured table as its justification comment; record losing numbers with "do
+not switch without re-measuring".
+
+The A/B vehicle for ANY single-GPU llama.cpp knob (batch/ubatch, KV type,
+slot split, quant rung — not just spec flags) is the transient systemd
+drop-in procedure in [[frank-large-benchmark]] and [[frank-small-benchmark]].
+
+## Is it worth it? (arithmetic before deployment)
+
+Answer against the INCUMBENT, not against zero.
+
+**Step 0 — check whether the target already speculates.** Many engines ship a
+drafter of their own; then your baseline is `A_incumbent / t_s`, not `1 / t_s`.
+halogen-flash-server: MTP head ON by default — read its own accounting, e.g.
+`serve_api: mtp 63 tok in 1.31s = 47.97 t/s | 33 rounds, commit 1.88/round`.
+llama.cpp: `/metrics` counters and per-request `draft acceptance` lines.
+
+**Step 1 — measure the two step times, never guess.** Target step `t_s` = 1 /
+(client-visible tok/s ÷ commit-per-round); ask for the rate the operator
+actually SEES (engine-internal decode-only numbers run ~1.7–2x optimistic).
+Drafter step `d_s` from llama.cpp's per-task `eval time` lines when it is a
+plain model on the same box.
+
+**Step 2 — arithmetic:**
+
+```
+round_time = K * d_s + 1.1 * t_s        # K drafts proposed, K+1 verified in ONE target pass
+A          = (1 - p^(K+1)) / (1 - p)    # expected accepted tokens, p = per-position acceptance
+rate       = A / round_time             # compare to the incumbent
+break-even : A > rate_incumbent * (K*d_s + 1.1*t_s)
+```
+
+Worked example that decided a real case: `t_s = 67 ms` (25 tok/s, 1.67
+commit/round), `d_s = 12 ms` (27B IQ3_S drafter on the 3090) → break-even
+`p ≈ 0.70` at K=3, `≈0.68` at K=5; a realistic cross-finetune `p = 0.4–0.6`
+gives 17–20 tok/s, i.e. a LOSS. A drafter only ~3x faster per token cannot
+pay for K sequential forwards.
+
+**Step 3 — classify the drafter by per-round cost:**
+
+| drafter | cost per round | needs |
+|---|---|---|
+| plain model (`-md`/`draft-simple`) | **K sequential forwards** | 10–50x faster than the target |
+| n-gram / prompt-lookup (`ngram-*`) | ~0, no model | repeated context (agent edit loops) |
+| trained head (`draft-mtp`, EAGLE/DFlash) | **~1 pass** | the head must exist FOR THAT CHECKPOINT |
+
+Trained heads are target-specific: EAGLE/DFlash consume the TARGET's hidden
+states, so a dense-27B DFlash2 is useless for a 512-expert MoE trunk.
+
+**Step 4 — acceptance reality checks.** Fleet datum: a trained z-lab DFlash2
+at depth 7 accepted ~41 % on its matched target (`draft-dflash 7 → 43.5
+tok/s`); treat 40–60 % as the honest band for a matched pair. Cheapest proxy
+for `p` at temp 0 when both models are greedy and share a tokenizer: greedy
+continuation agreement on identical prompts, two API calls, no weights moved.
+llama.cpp can also synthesize the projection without any draft artifact:
+`--spec-synth-rates P0,P1,...` / `--spec-synth-len L`. Speculation pays IN
+PROPORTION TO HOW SLOW THE TARGET IS — the same 41 % gave 2.3x on an
+18.8 tok/s layer-split target and a loss on a 25 tok/s one; fast targets are
+usually sparse MoE, the worst case for drafting.
+
+**Step 5 — check the attachment surface BEFORE the arithmetic.** Closed
+engines refuse external drafters by design: halogen-flash-server's
+`DRAFTER_CAP` gates `dflash2` on the loaded checkpoint and an unavailable
+name is a deliberate 400 (a silent downgrade would make comparisons lie);
+there is no device routing in its loopback token protocol, so a drafter can
+never sit on another card. Pure llama.cpp DOES route the draft:
+`--spec-draft-device/-devd` (default follows `--device`), `-ngld`,
+`-otd`, `-ctkd/-ctvd`, `--spec-draft-n-max` (default 3), `--spec-draft-p-min`.
+One process sees only the devices its backends were built with — CUDA+Vulkan
+is the combination that shows a CUDA card AND an AMD iGPU (HIP+CUDA cannot
+coexist); a CUDA-only build has no `Vulkan0` to draft with.
+
+**Step 6 — cross-model pairs need a byte-identical tokenizer.** Speculation
+compares token IDs, so draft and target must map id → same string. Verify
+cheaply from the GGUF headers (first ~8 MB over HTTP, no full download):
+parse `tokenizer.ggml.tokens`, print vocab size, merges, `pre`, and sha256
+over the first 2000 tokens —
+
+```
+vocab=248320 merges=247587 model=gpt2 pre=qwen35
+sha256(first 2000 tokens)=8619c68fe4466d3a   # identical => ids are interchangeable
+```
+
+Identity is necessary, NOT sufficient — the drafter still has to be cheap and
+accepted. Report in this order: does the target already draft; break-even
+`p` at K=1,3,7 versus the incumbent; the realistic band and resulting rate;
+whether the drafter fits (VRAM) and the attachment surface exists; the one
+measurement that would settle it.
+
 ## Notes
 
 - Speculative decoding helps low-QPS, memory-bound, latency-focused workloads. It helps less at high QPS.
@@ -225,4 +415,8 @@ History: before `--spec-default`, the box ran DSpark (`draft-dspark`, retired 20
 - [[Dankclaw System Specs]]
 - [[Dankserver System Specs]]
 - [[VLLM Overview]]
-- [[deployment]]
+- [[deployment]]
+- [[GGUF Compatibility]] — header parsing and custom tensor-type lock-in
+- [[Llama Cpp Serving Sizing]] — slot-pool and drafter KV sizing arithmetic
+- [[Llama Cpp Multi-Device Splits]] — where the drafter's KV actually lands
+- [[frank-large-benchmark]] and [[frank-small-benchmark]] — the single-GPU env drop-in A/B vehicle

serverAppConfigs-docs/LLM/Dankclaw System Specs.md

@@ -101,8 +101,8 @@ cable between claw and server is required for 10G.
 
 - Unified memory eliminates separate VRAM limits. The GPU uses system RAM.
 - The GPU budget is the GTT aperture: ~114 GiB (`gtt_total` in `/sys/class/drm/card1/device/`; `llama-server --list-devices` reports 116,732 MiB). rocm-smi's smaller SVM/dedicated carve-out is **not** the budget — the 2026-08 "64 GiB rocm-smi" lesson is stale. Do not use `free` or `dmidecode` for the GPU budget.
-- The active service is `frank-large` on port 7770: stock llama.cpp (the unsloth b11030 prebuilt) serving unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL (27B dense, q8_0 KV, DFlash2 drafter). See [[frank-large-unsloth-qwen38-27b-q8xl]].
-- The active runtime (the unsloth b11030 prebuilt) **bundles its own ROCm runtime** next to the binary (`RPATH=$ORIGIN/../lib`), so it has no host SDK gate — the system ROCm 7.2.4 is not on its path. The retired CIRU venv's TheRock 7.15 SDK (`LD_LIBRARY_PATH`) is no longer in the serving path.
+- The active service is `frank-large` on port 7770: **gufo** (gufo-org/gufo) v0.1.1, the Strix Halo inference engine, built from source at `/home/dank/GitPackages/gufo-v0.1.1/`, serving unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL (27B dense, DFlash2 drafter). See [[frank-large-gufo-qwen38-27b]].
+- The active runtime **links against the host ROCm** (`/opt/rocm` → `/etc/alternatives/rocm` → 7.2.4) through its RPATH, so it needs the host HIP stack rather than a bundled one — the opposite of the retired llama.cpp prebuilt, which carried its own. Building it needs the ROCm dev packages that are installed, and two workarounds recorded in the runbook (drop `librocwmma-dev`; pin `--gcc-install-dir` to GCC 15). The retired CIRU venv's TheRock 7.15 SDK (`LD_LIBRARY_PATH`) is not in the serving path.
 - The retired vLLM venvs (`/home/dank/vllm-venv`, `/home/dank/vllm-vk-venv`) were moved to `/home/dank/.AGENTTRASH/` on 2026-08-18.
 - 1.1 TB free disk space is ample for model storage.
 

serverAppConfigs-docs/LLM/Dankserver System Specs.md

@@ -63,16 +63,16 @@ apt lists remain on disk but no NVIDIA PCI device exists.
 
 | Property | Value |
 |----------|-------|
-| System RAM | 60 GiB |
-| Swap | 63 GiB |
+| System RAM | **246 GiB** (upgraded for the Lucebox DeepSeek deployment of 2026-09-26; this page said 60 GiB until 2026-09-28 — a stale figure that survived the board swap) |
+| Swap | 63 GiB (`/swapfile`) |
 | /dev/shm | 32 GiB |
 
 ### Storage
 
 | Filesystem | Size | Used | Avail | Use% |
 |------------|------|------|-------|------|
-| / (nvme0n1p2) | 915G | ~640G | ~280G | ~70% |
-| /mnt/entropy (dm-0) | 91T | 86T | 910G | 99% |
+| / (nvme0n1p2) | 1.8T | 927G | ~822G | ~54% |
+| /mnt/entropy (dm-0) | 91T | 87T | **244G** | **100%** — full; do not plan model payload here |
 
 Models live under `/home/dank/models`; the serving build tree under
 `/home/dank/GitPackages/llama.cpp-dflash2` (the older `charlie-ROCmFPX` tree is

serverAppConfigs-docs/LLM/FastFlowLM NPU Inference.md

@@ -271,6 +271,83 @@ graph TD
 6. Step up in size: 9B, then gpt-oss-20b.
 7. Attempt `qwen3.6-moe:35b-a3b` only after the smaller runs succeed, and only with the production service stopped. Watch `free -h` during the model download and load.
 
+## Registering packs absent from the catalog (NPU2)
+
+Use when adding a non-catalog NPU2 pack (an extraction or fine-tuned model)
+to the FLM registry. Registry facts: user registry
+`/home/dank/.config/flm/model_list.json`, shape
+`{"model_path": "models", "models": {"<tag>": {"<size>": {name, url,
+file_url, size, flm_min_version, files[], vlm, default_context_length,
+max_prefill_len, details{format:"NPU2", family, …}, label[]}}}}`; serve
+resolves tags through `FLM_CONFIG_PATH` at startup, and `flm list` with the
+env var set lists custom tags (`⏬` = not downloaded).
+
+1. Back up the registry JSON first (`cp -a`).
+2. Insert the entry via a python json edit + atomic `os.replace`. Set
+   `files` to the FULL HF file list — **including `vision_weight.q4nx` even
+   for text-only packs**: the qwen3.5-family loader opens it unconditionally
+   and a missing file makes the first request HANG silently (no error log).
+3. **Do not use `flm pull <tag>` for custom entries** — it throws
+   `json.exception out_of_range key '<tag>:<size>' not found` while building
+   the download list. Place files manually into
+   `/home/dank/.config/flm/models/<PackName>/` with
+   `curl -sfL "https://huggingface.co/<repo>/resolve/main/<file>?download=true"`
+   — plain `raw/main/<file>` returns 135-byte LFS pointer stubs, not
+   weights; verify byte counts against the HF tree API before trusting a
+   download.
+4. Kernel packs: copy from the same-shape family pack (shape-match
+   `config.json` on hidden_size/num_hidden_layers/head_dim first) into
+   `/opt/fastflowlm/share/flm/xclbins/<PackName>/` via `sudo cp -a`
+   (root-owned dir).
+5. Deploy/restart the serve unit (the deploy path, not ad-hoc restarts) so
+   the new process loads with all files present.
+6. Verify with a REAL completion request (`model: "<tag>:<size>"`) —
+   `/v1/models` alone proves registration, not loadability.
+
+Hard constraints: only ONE `flm serve` process can hold the NPU hardware
+context — a second concurrent serve process fails every load with
+`DRM_IOCTL_AMDXDNA_CREATE_HWCTX err=-22` (aie2 mgmt msg opcode 0x2 status
+0x2000003); never run scratch/verify instances while production holds the
+NPU. Tag switching happens INSIDE one serve process; the first request
+after a model switch pays ~6–11 s cold-load cost (visible in
+`usage.load_duration`). Over ssh, never `pkill -f "<pattern>"` with a
+pattern present in your own command line — use bracket-guarded patterns
+(`[q]wen3.5:4b`) or exact PIDs.
+
+## Tag load-failure diagnosis (200 with an error body)
+
+`POST /v1/chat/completions` answers **HTTP 200** with a flat
+`{"error":"Failed to load <tag> model!"}` and no `choices`; `/v1/completions`
+returns a nested OpenAI-shaped `server_error` (code 500) also under HTTP
+200. A status-code-only client reads this as a successful empty completion —
+**always check for the `error` key**. Two distinct causes, distinguished via
+`journalctl -u <flm-service> --full`:
+
+1. **Unknown tag** (not in the registry): immediate error, nothing loaded.
+   Correct behavior, not a bug.
+2. **Registered tag whose kernel pack is missing**: FLM resolves the tag,
+   downloads missing weight files on the fly (first request hangs ~90–110 s),
+   then fails `Failed to load model: No such file
+   '/home/dank/.config/flm/xclbins/<PackName>/<kernel>.xclbin'`; subsequent
+   attempts fail fast (~2 s). Root cause: the serve script exports
+   `FLM_XCLBIN_PATH=/home/dank/.config/flm/xclbins` (user dir), so every
+   kernel lookup resolves under the user dir, which only contains symlinks
+   someone linked. **A tag listed in `/v1/models` is NOT proof it is
+   loadable on this host.** Fix: one symlink per missing pack
+   (`ln -s /opt/fastflowlm/share/flm/xclbins/<PackName>
+   /home/dank/.config/flm/xclbins/<PackName>`) — verified live: after the
+   symlink the first request loaded and answered in 7.4 s wall.
+
+Switch-cost behavior: the `model` field routes per-request (no restart
+needed); a request whose model differs from the resident one triggers a full
+reload (7–11 s wall for an 8-token completion on ~4B-class packs);
+alternating tags pays the swap each time. Single-lane NPU: concurrent
+requests queue FIFO behind the current lock holder (`NPU Locked!` →
+`NPU Lock Released!`). Probe recipe: `/v1/models` for registered tags; grep
+the journal for `Failed to load model` (the quoted xclbin path names the
+missing pack dir exactly); diff `ls ~/.config/flm/xclbins/` (symlinks =
+installed) against `/opt/fastflowlm/share/flm/xclbins/` (available).
+
 ## Incident: qwen3.6-moe unreachable (2026-08-16)
 
 Symptom: a client on dankware could not reach `qwen3.6-moe:35b-a3b` on port 52625. The journal shows three separate causes in the 14:19-15:03 window.

serverAppConfigs-docs/LLM/FreeToken Builds.md

@@ -0,0 +1,97 @@
+---
+type: reference
+title: FreeToken Builds
+description: Building FreeToken (the official franksdank fork of FreeTok Tokenworks' freetoken vLLM-MTP runtime) on Dankclaw — the CUDA vLLM venv build and the ROCm build, with the dependency traps each one hit.
+tags: [freetoken, build, dankclaw, vllm, mtp]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# FreeToken Builds
+
+FreeToken = the official franksdank fork of FreeTok Tokenworks' freetoken
+vLLM-MTP runtime. Two build paths exist, both on **Dankclaw**, with
+different dependency traps. Both were verified end-to-end 2026-09-11/12.
+
+## Common verification (do not skip)
+
+**Verify the build by real completion, not by version string.** A version
+bump alone does not move data through the model — a build can print a new
+version and still not have completed a real token. Before declaring a
+FreeToken build deployable, run an actual completion (any tiny prompt) and
+read the returned content:
+
+```python
+import json, urllib.request
+req = urllib.request.Request("http://<host>:<port>/v1/chat/completions",
+    json.dumps({"model": "frank-small", "messages": [{"role": "user",
+    "content": "Hi"}]}).encode(), {"Content-Type": "application/json"})
+print(json.load(urllib.request.urlopen(req, timeout=600)))
+```
+
+A 200 with a real `choices[0].message.content` is the only build green.
+
+## CUDA build (the torch/vLLM dependency trap)
+
+Both paths build on **Dankclaw** with the same preinstall state: Python
+3.12.10 in the system venv and **Torch 2.10.0.dev** (nightly, repo
+download.pytorch.org/whl/nightly/cu130, installed with
+`uv pip install --index-strategy unsafe-best-match "torch==2.10.0.dev<date>+cu130"
+--index-url https://download.pytorch.org/whl/nightly/cu130` — use
+`--index-strategy unsafe-best-match`, `uv` otherwise refuses the dev
+version). Verify the exact pair first; mismatched preinstalled deps are the
+#1 build-failure cause.
+
+Then install the **matching vllm dev wheel**:
+
+```bash
+uv pip install --index-strategy unsafe-best-match \
+  --index-url https://wheels.vllm.ai/<commit-sha>/vllm-0.6.6.dev0+g<shortsha>.cu130-cp312-cp312-linux_x86_64.whl \
+  --extra-index-url https://wheels.vllm.ai/nightly/cu130
+```
+
+The wheel URL pattern is fixed:
+`https://wheels.vllm.ai/<full-commit>/vllm-0.6.6.dev0+g<8-char>.cu130-cp312-cp312-linux_x86_64.whl`
+— read it back from vLLM's CI-run page (`run details →
+build_and_install_package` job, "install the vllm wheel" step). **The trap:
+vLLM refuses versions < its install_requires floor at runtime, so the wheel
+version must be ≥ the runtime's floor.** A wheel built at a lower base
+(0.5.x) downloads but `import vllm` fails with a version constraint error —
+that failure looks like a build failure and is not one; upgrade the wheel,
+not the code.
+
+If `uv pip install -e . --no-deps` refuses (some repos pin the
+`--no-deps` branch off), fall back to `python3 setup.py` inside the venv,
+after `python -m venv --copies .venv && source .venv/bin/activate` and
+`uv pip install --no-build-isolation "torch==…"` — the venv must NOT
+re-resolve torch (a second torch from PyPI d1 breaks the CUDA extension ABI).
+
+## ROCm build (gfx1151 via TheRock nightly)
+
+Same venv discipline, ROCm wheels from TheRock's nightly index:
+
+```bash
+uv pip install --index-strategy unsafe-best-match \
+  --index-url https://rocm.nightlies.amd.com/v2/gfx1151-dcgpu-supports_fbsgens_split_0717-rc1/ \
+  "torch==2.10.0.dev<date>+rocm" "triton==3.6.0" "pytorch-triton==3.6.0"
+```
+
+Then `vllm` itself (`uv pip install -e . --no-build-isolation --no-deps`,
+repo branch `specs-mtp-dev`, commit `2ee440d`), after copying the two
+prebuilt extension wheels into the tree (the repo's `setup.py` skips
+compiling them when found). The ROCm build also needs
+`install-fastrak-gfx1151.sh` from the repo's scripts and
+`FASTRACK_HOME=/home/dank/FastTrack-runtime` in the unit env. FreeToken's
+x86_64-gfx1151 prebuilt extension (`freetoken-ops`) ships on HF hub
+`franksdank/freetoken-ops-gfx1151` — fetch it once and point
+`FREETOKEN_OPS_HOME` at it; never rebuild the extension from source unless
+the extension API changed.
+
+`--index-strategy unsafe-best-match` is load-bearing on both paths: the
+nightly indexes carry version pairs that `uv`'s strict resolver otherwise
+rejects, and the error (`No candidate found for pytorch-triton`) reads like
+a missing package.
+
+## Related
+
+- [[Model Swap and Rollback]] — the swap/deploy story these builds serve
+- [[Llama Cpp Serving Sizing]] — the harness measurement discipline

serverAppConfigs-docs/LLM/GGUF Compatibility.md

@@ -0,0 +1,123 @@
+---
+type: reference
+title: GGUF Compatibility and Runtime Lock-in
+description: How to answer "will engine X load this GGUF?" from the file's header alone, why a custom tensor type forces the runtime, and the discipline for pinning a fork build that can actually serve the file.
+tags: [gguf, llama.cpp, forks, quantization, speculative]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# GGUF Compatibility and Runtime Lock-in
+
+**The law: a GGUF's tensor-type histogram and ftype pick the runtime — read the
+header before promising a file to any loader.** Stock llama.cpp, Ollama and LM
+Studio refuse custom types at load (`unknown tensor type`); a fork-private type
+means exactly one engine can serve the file. Measured on this fleet repeatedly:
+`Qwen3.8-27B-Q8_0_ROCMFPX_AGENT.gguf` is ftype 115 with 311 tensors of type 103
+(`GGML_TYPE_Q8_0_ROCMFPX`) — only charlie12345/ROCmFPX reads it. The full type
+tables live in [[Llama Cpp Runtimes]] and [[Llama Cpp Code Diff]]
+(ggml_type 100–108, ftype families 100–106 / 110–113 / 118).
+
+## 1. Read the header, do not guess
+
+Parse the GGUF KV section and histogram the tensor types plus
+`general.architecture.file_type` — one python read over the first few MiB, no
+model load. Counting types out of the KV section settles a runtime argument in
+minutes; repo names and model cards do not.
+
+## 2. Compare routing variants by tensor mix, not by name
+
+Two files of the same quant can differ in how many tensors stay at plain Q8_0
+(measured: 187 vs 2 output-head tensors in an `_AGENT` variant of the same
+size). That difference is what moves spec-decode acceptance (0.953 vs 0.911)
+— the "same size" file is not the same file.
+
+## 3. Pin the build by commit SHA
+
+A fork that can read the file today can still crash mid-load: the 2026-08-25
+ROCmFPX builds read the ftype-115 file and then SIGSEGV in
+`ggml_compute_forward_set_rows`. Record the SHA in the build script
+(`scripts/build-rocmfpx.sh` pins `36e9acd`) and check it out explicitly — never
+a branch.
+
+## 4. Header diff method (candidate vs known-good)
+
+Answer "will engine X load this GGUF?" by diffing the candidate's header
+against a file the engine **already loads** — never by filenames or model cards;
+filenames lie about quantization and about MTP/vision heads.
+
+1. **Get the bytes**: both files, from byte 0, at least 64 MiB. Tokenizer KVs
+   (151k token strings + merges) routinely eat 5–8 MiB, so a small range
+   desynchronises the parser. True size from `curl -sIL` → `x-linked-size`
+   (a bare HEAD `Content-Length` can lie):
+
+   ```sh
+   curl -sL -r 0-67108863 -o cand.bin \
+     "https://huggingface.co/<owner>/<repo>/resolve/main/<file>.gguf"
+   ssh <host> 'head -c 67108864 /path/to/known-good.gguf' > good.bin
+   ```
+
+2. **Parse — and lock the parser against the known-good file first.** Layout
+   (v2/v3): `magic[4]` `version u32` `n_tensors u64` `n_kv u64`, then KVs, then
+   tensor infos. Strings are u64 length + bytes. **The value-type enum is NOT
+   llama.cpp's `gguf.h`:** in the files seen (2026-09-28, unsloth + ajgazin UD
+   GGUFs) type **8 = STRING, 9 = ARRAY**; llama.cpp's header has STRING=2,
+   ARRAY=3 and desynchronises immediately. Working table:
+
+   ```
+   0 u8   1 i8   2 u16  3 i16  4 u32  5 i32  6 f32
+   7 bool 8 STR  9 ARRAY(u32 et, u64 n)  10 u64  11 i64  12 f64
+   ```
+
+   Validate before trusting: the parse is right iff the tensor-info block ends
+   somewhere sane and names look real (`blk.0.attn_q.weight`,
+   `token_embd.weight`, `output_norm.weight`). EOF throw = wrong enum or
+   offset — do not work around it.
+
+3. **Diff what engines actually bind on**: `general.architecture`; tensor
+   count; tensor name list in order; per-tensor `ggml_type` and dims; geometry
+   KVs (`<arch>.block_count`, `.embedding_length`, `.feed_forward_length`,
+   `.attention.head_count[_kv]`, `.rope.*`, `.ssm.*`, `.context_length`,
+   `.nextn_predict_layers`); MTP tensors (`nextn` in the name); vision tensors
+   (`visual`, `v_` prefixes). A same-family fine-tune quantized by the same
+   tool usually comes out byte-identical in every one of these — that is the
+   answer.
+
+4. **Then read the engine's source before calling a delta a blocker.** A
+   metadata difference only matters if the engine reads that key
+   (`grep -rn "file_type\|recurrent_layers\|full_attention_interval" <src>/`).
+   Worked example (gufo v0.1.1): `general.file_type` is deliberately not
+   consulted (UD-*_K_XL ships a stale ftype 15; the tensor table rules), and
+   `qwen35.attention.recurrent_layers` is never read (layer kinds come from
+   `full_attention_interval`) — both deltas were non-issues the header diff
+   alone would have flagged wrongly.
+
+5. **Say what you did not do.** Header analysis is strong evidence, not a
+   load. State byte sizes, check free disk before proposing a fetch, and do
+   not download tens of GB unasked — pointing the engine at the file is the
+   definitive test.
+
+**Report shape:** verdict first (loads / will not / loads-with-caveats), then
+the side-by-side diff table, then only the load-bearing caveats: text-only vs
+vision tensors, chat-template differences (matters for llama.cpp, not for
+engines with compiled-in Jinja), license, disk headroom, and whether the base
+model is abliterated/uncensored.
+
+## 5. A/B a fork's spec decode by stripping the whole flag set
+
+`--model-draft` without `--spec-type` segfaults at load (measured 2026-10-02,
+core dump, no diagnostic). A baseline "no-draft" arm must remove `--model-draft`
+AND every `--spec-draft-*` AND `--spec-type` together — the flags move as a set.
+
+## 6. Before deleting a retired payload, read the rollback tag's env file
+
+A rollback may need a small companion dir: a drafter living in a directory
+named after an old quant looks like trash but is load-bearing. Same class as
+the gufo payload removal of 2026-10-03, which turned that rollback into a
+re-download.
+
+## Related
+
+- [[Llama Cpp Runtimes]] — which fork ships which custom types
+- [[Llama Cpp Code Diff]] — the enum tables and code-level differences
+- [[frank-large-rocmfpx-q8agent-mtp]] — the ftype-115 lock-in in production
+- [[Llama Cpp Serving Sizing]] — pricing what a payload costs once it loads

serverAppConfigs-docs/LLM/Heterogeneous Split Triage.md

@@ -0,0 +1,159 @@
+---
+type: reference
+title: Heterogeneous Split Triage
+description: Triage before building when someone asks to split a model across unequal accelerators on one host (24 GB GDDR6X + LPDDR APU, PCIe x4 link) — link characterisation, the real bottleneck, what placement never buys, UMA pools, drafter splits, and the verdict ranking.
+tags: [heterogeneous, multi-device, llama-cpp, triage, pcie]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Heterogeneous Split Triage
+
+Produced the 2026-09-16 no-go verdict on splitting Qwen3.8-Flash-Next across
+the RTX 3090 + Strix Halo pair. Use it to answer "could we split model X
+across device A and device B" with arithmetic instead of enthusiasm.
+
+## 1. Characterise the link (bytes vs latency)
+
+```bash
+lspci -vv -s <bdf> | grep -E "LnkCap:|LnkSta:"          # capability vs actual
+nvidia-smi --query-gpu=pcie.link.gen.max,pcie.link.gen.current,pcie.link.width.current --format=csv
+```
+
+Expect **idle-downshift**: a Gen4 x4 dock idles at Gen1 x4 (~1 GB/s) and
+upshifts under load. Measure while the device is busy, or you size the design
+against the wrong number. Per-hop cost against the hidden state
+(`n_embd × 2 B` for f16): a 5120-dim model moves 10 KiB/hop ≈ 1.3 µs at
+Gen4 x4.
+
+- **Bytes per token ≈ hops/token × hidden state** — typically ~1 MiB/token
+  for 2 hops/layer × 64 layers ≈ 0.3 % of a 40 ms token. **Bytes are almost
+  never the constraint.**
+- The real budget line is **hop COUNT**: every device boundary is a
+  host-side sync. A runtime that splits per *op* hits ~100 splits/token and
+  pins the GPU at ~5 % SM. Measure `graph splits` in the loader log; never
+  assume.
+- **Design rule: partition by where the weights live, never by where the
+  activations flow.** Resident weights, activations crossing.
+
+## 2. Get the workload's real bottleneck (KV is usually it)
+
+```
+KV/token = kv_bytes_per_position × context
+```
+
+Attention reads the whole cached context every token, so KV/token scales
+with context — compare it against expert/weight traffic per token and total
+token time. At long context on a hybrid arch (few attention layers, many
+positions), **KV dominates**. Consequences:
+
+- The highest-leverage placement is **KV cache on the fastest-memory device,
+  together with its attention compute** — never KV alone: separating them
+  puts KV/token over the link, which is fatal (GBs across 7.88 GB/s =
+  hundreds of ms/token).
+- Check the fit: `kv_positions_per_slot × slots × bytes_per_position +
+  attention_trunk ≤ VRAM`.
+
+## 3. What placement never buys
+
+Prefill is compute-bound and expert weights are read once per batch, so
+moving them to faster memory buys nothing while adding a serial stage.
+Expect a regression, and size it from calibration: on Strix Halo + RTX 3090
+(Qwen3.8-Flash-Next, ik_llama.cpp) moving 12.5 % of the expert mass to the
+GPU gave **decode +12 %, prefill 192 → 32 tok/s**. A
+`1/(f_big/BW_slow + f_fast/BW_fast)` harmonic calculation reproduces that
+decode gain to within ~2 points — usable for decode projections, and it
+assumes sequential stages (true at batch 1; no pipelining).
+
+## 4. Is the second device even a separate pool?
+
+On a UMA APU, check before assuming additivity:
+
+```bash
+cat /sys/class/drm/card*/device/mem_info_vram_total   # BIOS carve-out (small)
+cat /sys/class/drm/card*/device/mem_info_gtt_total    # aperture onto SYSTEM RAM
+rocm-smi --showmeminfo all | grep -i gtt              # used == what the CPU also sees
+```
+
+A large GTT window means the "GPU memory" is the *same* LPDDR under one
+memory controller: adding a stage there adds a compute engine, not capacity
+or bandwidth. Only the discrete card's VRAM is additive. Fencing more
+(`amdgpu.gttsize`, BIOS UMA size) just removes bytes from CPU-visible RAM.
+
+## 5. Speculative decoding as the "cheap" split
+
+A drafter on the *other* device is bandwidth-free (token ids per round, tens
+of bytes), so it is tempting. Before proposing it:
+
+- the drafter needs the target's **exact tokenizer** for a plain draft model
+  (compare GGUF `tokenizer.ggml.tokens`, never trust the family name), or it
+  must be a head **trained for that target** (EAGLE/DFlash/MTP are
+  conditioned on the target's hidden states and are not portable between
+  models);
+- a plain drafter costs **K sequential forwards per round**, so it must be
+  10–50× faster than the target. Break-even with `r = t_draft/t_target`:
+  `rate_ratio = A/(K·r + 1)`, `A = (1−p^(K+1))/(1−p)`; at `r ≈ 0.4–0.5`
+  (a drafter only ~2× faster) break-even needs `p ≈ 0.7+` at K=3;
+- compare against the **incumbent**, not the no-draft baseline: a target
+  already running its own MTP head is a high bar;
+- check the runtime accepts an external drafter at all — closed engines may
+  enumerate a fixed drafter set and 400 anything else (capability gates
+  keyed on the loaded checkpoint). That is a design boundary, not a config
+  gap. In llama.cpp the portable mechanism is `--spec-draft-device/-devd`
+  (draft gets its own device list) plus `-ngld`/`-otd`/`-ctkd -ctvd`.
+
+## 6. Expressing a split (llama.cpp, the usual vehicle)
+
+```bash
+-dev CUDA0,Vulkan0 -sm layer -ts F0,F1        # contiguous layer ranges (simplest)
+-ot "\.attn_.*=CUDA0" -ot "\.exps\.=Vulkan0"   # per-tensor, by ROLE (KV co-location)
+```
+
+- buffer types match by name at load (`ggml_backend_buft_name`), so device
+  names work as `-ot` values;
+- `-cmoe`/`-ncmoe` are **CPU-only**; device-side MoE placement is `-ot`;
+- several backends in one binary is fine (CUDA + Vulkan is the practical
+  pair; HIP + CUDA do not coexist) — `GPU_BACKEND_A=OFF` in the existing
+  build's CMakeCache tells you what needs adding;
+- contiguous ranges are the WRONG split when KV bandwidth dominates: they
+  drag half the attention layers (and their KV) back to the slow pool.
+
+## 7. Hazards checklist
+
+- Paged/sidecar tables (huge n-gram/embedding tables) must stay where they
+  are read from; moving them is pure loss.
+- Keep `mmap` on for models with huge embedding tables (anonymous memory
+  gets the process cgroup-killed).
+- CUDA graph capture can abort on unusual graph shapes (hyper-connections)
+  — check before assuming optimisations apply.
+- A brand-new backend on the second device may not implement the arch's ops
+  at all: make the first measurement a **baseline of the runtime on that
+  device alone** — a split's arithmetic is worthless if the baseline is
+  worse than the incumbent engine.
+- A split consumes the fast device entirely — check what else it was
+  serving before trading it away.
+- Benchmark discipline on production hosts: stop and mask first
+  ([[Fleet Ops Playbook]]).
+
+## 8. Verdict shape
+
+Rank and report; prefer the boring answer when it wins:
+
+1. **Two independent endpoints** (zero shared traffic) — often the real
+   answer to "what's a useful combination", especially over a narrow link.
+2. **One process, weights-resident, KV co-located with attention on the
+   fast pool** — promising and quantifiable (+% decode at long context);
+   needs a dual-backend build and a baseline measurement.
+3. **Hot-expert cache / per-token weight streaming** — affordable only if
+   the miss set per token is small (a few experts; check
+   `misses × bytes × tok/s` against link bandwidth) and overlapped; injects
+   PCIe latency into the critical path; no mainstream runtime implements a
+   dynamic expert cache.
+4. List explicitly what does NOT help: TP/collectives across unequal
+   devices, KV on the far device, model loading over the link,
+   prefill-motivated placements.
+
+## Related
+
+- [[PCIe Link and Atomics]] — the link and collective walls underneath
+- [[Llama Cpp Multi-Device Splits]] — the deployed split shapes
+- [[Llama Cpp Serving Sizing]] — KV and pool arithmetic

serverAppConfigs-docs/LLM/LLM Bench Windows and Measurement.md

@@ -0,0 +1,271 @@
+---
+type: reference
+title: LLM Bench Windows and Measurement
+description: Stopping-and-restoring benchmark windows on the fleet LLM hosts (llama-bench sweeps, HTTP param benches, multi-arm vLLM offload sweeps), honest benchmark methodology, and the usage-counter measurement law under speculative decoding.
+tags: [benchmarking, llama-bench, ab-testing, speculative, vllm, fleet]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# LLM Bench Windows and Measurement
+
+Fleet law first: **STOP the production service before any GPU experiment**
+(AGENTS.md Benchmarking Policy) — a bench beside a live service starves the
+GPU and corrupts results; an unisolated GPU experiment has precedent for
+wedging a host until power cycle. Reference harnesses:
+`scripts/bench-batch-ubatch.sh`, `scripts/autoresearch.sh`,
+`scripts/autoresearch/frank-primary-arms.sh`.
+
+## 1. The window structure (one config per arm, a driver owns the window)
+
+1. **Stop + gate before every arm** — `wait_for_clear`: unit NOT active AND
+   no RUNNING container of that name (`podman ps`, not `-a`; `-a` counts
+   EXITED containers and can never clear — `rm -f` the exited one first) AND
+   port free (`ss -ltn`). The graceful-stop race cost a whole arm 2026-10-01:
+   `ExecStop=podman stop -t 120` drains up to 2 min; a driver that sleeps 3 s
+   then polls `/health` gets **200 from the still-running PRODUCTION engine**
+   — the arm "passes" with production's boot log, then `podman exec` dies
+   with `container state improper`.
+2. **Boot the arm as the service, not a scratch port** (no scratch ports on
+   LLM hosts — one card, one server). The serve script takes its config via
+   env-file override (`FRANK_SMALL_ENV_FILE`, `FRANK_PRIMARY_AR_ENV_FILE`)
+   pointing at a staged copy; the checkout is never mutated.
+3. **Fast-fail the health wait**: poll `podman ps | grep -qx <name>` inside
+   the loop — a refused boot dies in one poll instead of 20 min. Capture the
+   serve script's FULL stdout on failure (a `tail -8` loses engine-init root
+   causes; raise happens above the tail).
+4. **Measure** (§4), capture boot-log engine facts (KV pool, arena slots,
+   SpecDecoding lines) — they survive measure crashes.
+5. **Teardown + converge**: kill by explicit PID/pgid (never pattern kills),
+   `podman rm -f` scratch, wait `MemAvailable ≥ ~230 GiB` on dankserver
+   before the next boot; restore production from the EXIT trap and verify the
+   **conjunction oracle**: health 200 + `is-active` + `is-enabled` +
+   mask-symlink clean. A bare `systemctl start` is not a restore oracle.
+   (`systemctl mask` FAILS on a relinked unit — log it, explicit stop is
+   enough: systemd never auto-restarts an explicitly stopped unit.)
+6. **Warm boots cycle ~2 min** (page cache), cold ~8 min — size timeouts for
+   mixed speeds.
+
+**Destructive tails (2026-10-01):** a scratch script whose tail does
+`systemctl stop` + `podman rm -f <PRODUCTION container>` fires later if a
+probe loop survived the driver kill — it can destroy a just-deployed
+production engine (signature: deploy exits green, then seconds-to-minutes
+later `Exited (0)`, `Deactivated successfully`, clean exit = removed, not
+crashed). Before ANY deploy: `pgrep -af` every scratch script of the session
+and kill strays by explicit PID; killing the dev-side job does NOT kill the
+host-side `deploy-remote.sh` (two of them once managed the same unit).
+
+Two failure shapes inside the window that read as "the bench died":
+
+- **Stale reads from a stopping container.** `podman stop` answers while the
+  engine inside is still flushing; a `podman logs` taken in that window
+  returns the boot-prefix only. Poll `podman ps` for `Exited (0)` (or
+  `podman wait`) before reading logs, or you will write "no results" into
+  the table for a run that has them.
+- **A D-state worker wedges the stop.** A bench child that wedged in the
+  kernel ignores SIGTERM, so `podman stop -t N` burns its full timeout and
+  then the container is `Stopping` forever. Kill by container first
+  (`podman kill <ctr>` → SIGKILL path), verify the process is gone from
+  `podman ps` AND `nvidia-smi`, and only then re-serve. Never let the
+  window script proceed to a redeploy with a zombie bench worker holding
+  VRAM.
+
+## 2. llama-bench sweeps (offline batch/ubatch/split/KV/depth)
+
+**What llama-bench cannot express — decide BEFORE designing:** no
+`--ctx-size` (`n_ctx = n_prompt + n_gen + n_depth`, so a bench can NEVER prove
+activation buffers fit beside the DEPLOYED KV — only the real unit answers
+that), no `--parallel` (no slot shape, no concurrency), no `--spec-*`
+(`tg` numbers are the drafterless baseline, useful as a control only), and a
+warmup run is itself a full-size run. `-d N` is a depth PREFIX (paid once per
+cell via a `llama_state_seq_*` save/restore); without it a `tg` run is at
+~zero depth and meaningless for long conversations.
+
+**The separator trap:** `/` separates devices WITHIN one configuration, `,`
+separates sweep values — for both `-dev` and `-ts`. `-ts 36,29` parses as the
+single value 36 → everything on one card. `--tensor-split 36,29` (server) →
+`-ts 36/29` (bench). Ranges exist: `-ub 512-2048+256`.
+
+**Harness shape:** one invocation per cell (an infeasible cell calls
+`exit(1)` and kills every later cell of a cross-product invocation — and
+infeasibility is usually the question); TSV row per cell as it completes +
+`bench.json` (`-o json`) + `bench.log` (`--progress`) + a 2 s `rocm-smi
+--showmeminfo vram` VRAM trace per cell (a cell that allocated less than
+intended looks fast); resume by skipping non-empty cell outputs; stop the
+service in the harness, restore from `trap … EXIT`, verify afterwards. Match
+the deployment's load flags exactly (`-ngl -sm -ts -fa -ctk/-ctv -t`).
+
+**Run under a transient unit, not a shell:**
+
+```bash
+sudo systemd-run --unit=<name> --collect --no-block --property=Type=oneshot \
+     --property=LimitMEMLOCK=infinity --property=User=dank \
+     /home/dank/serverAppConfigs/scripts/<harness>.sh
+```
+
+Deploy units set `LimitMEMLOCK=infinity` for ROCm staging; an SSH shell has
+8 MiB and fails the load for an unrelated reason. This also survives ssh
+disconnects. **Boot rungs under the production environment, not a bench
+shell** — a rootless shell probe booted 3×262144 bf16 that production refused
+(systemd `User=` + memlock change what the pinned-expert path holds):
+`sudo systemd-run --uid=<user> --unit=<name> -p LimitMEMLOCK=infinity -p
+Restart=no env VARFILE=… /path/serve.sh`.
+
+**Verify before burning hours:** dry-run against a STUB binary with
+`SERVICE=<nonexistent-name>` (the trap/stop never reached), tiny grid,
+including one forced-`exit 3` cell — prove the TSV row, failure capture,
+resume logic. Validate the real flag set in ~1 s by pointing `-m` at a
+nonexistent file. **Reading:** bracket the sweep (reference cell first AND
+last); check per-cell `stddev_ts` (a difference smaller than spread is not a
+result); single-sequence decode is usually flat across batch/ubatch (prefill
+knob); batch ≥ ubatch (llama.cpp clamps ubatch to batch, silently duplicating
+smaller-batch cells); report what the grid cannot prove (drafter, slots, real
+context).
+
+## 3. HTTP benches of server-side parameters
+
+`llama-bench` benches the bench binary, not the running server — server-side
+params (`--threads-batch`, `--ubatch-size`, `--spec-type`, `--parallel`) need
+HTTP:
+
+- **Prefill**: large prompt, `max_tokens=1` (server crashes on 0), measure
+  `prompt_tokens / wall_time`.
+- **Decode**: `max_tokens=16384` minimum, NEVER 1 (warmup, cache growth and
+  spec acceptance all stabilize over thousands of tokens); measure
+  `predicted_per_second` from `timings` or `completion_tokens / wall_time`;
+  temperature 0.
+
+Known fleet harnesses: `scripts/bench-threads-batch-http.py`,
+`scripts/bench-decode-threads-batch.py`, `scripts/bench-spec-decode.py`.
+sglang-specific: `max_tokens=1` on `/v1/chat/completions` with the qwen3
+reasoning parser streams zero deltas — use `/v1/completions`.
+
+## 4. The measurement law (under speculative decoding this is absolute)
+
+**Never count SSE chunks.** Rejected drafts break the 1:1 chunk↔token
+mapping; chunk counting reported 8.3 tok/s for a config whose server counters
+said 23–26 — a 3x false regression that nearly abandoned the change that
+worked. Always:
+
+```python
+decode_tok_s = usage.completion_tokens / (wall - ttft)
+# stream=True, stream_options: {include_usage: true}; ttft = first content-bearing chunk
+```
+
+vLLM's 10 s `Accepted throughput` log lines are a rolling window, not a
+measurement (6.2 → 13.1 across adjacent windows on identical settings) — use
+them for acceptance STATISTICS only (`Avg Draft acceptance rate`, `Mean
+acceptance length`, per-position rates). Warm pass + ≥3 runs, median;
+temperature 0; ≥256 tokens. **A decode "jump" with early-stopped completions
+(ct < 256) and acceptance > 90 % is degenerate generation** (the fork's
+repetition-halt fired at ~140 tokens on corrupted numerics) — inspect text
+before believing a win. Run-to-run spread is 5–10 %: repeat arms; one sample
+nearly shipped the wrong setting.
+
+## 5. Honest methodology
+
+- **Match the corpus to the workload.** Token soup under-reports speculative
+  decoding (2.9–3.6 tok/update vs 3.86–5.8 on real code) and mis-ranks draft
+  settings; acceptance, not step time, moves throughput. Build the corpus
+  from the repo's own sources: `{"prompt": "<instruction>\n\n<file body>"}`,
+  `--custom-output-len`.
+- **Salt every run** with a unique per-arm token in the first bytes — it is
+  the prefix-cache key (and a proxy routing key). Without it arm 2 measures
+  arm 1's warm cache.
+- **Two workload shapes need two clients:** distinct prompts (batch
+  behaviour, prefill parallelism) and sequential turns with growing history
+  (TTFT of the first streamed chunk per turn — the only measurement that
+  exposes prefix-cache behaviour). These pull in opposite directions: a cold
+  batch spreads prefill across replicas, a conversation's next turn only pays
+  if it lands where its prefix is cached.
+- **Read the engine's ledger, not the card:** `Available KV cache memory` /
+  `GPU KV cache size: N tokens` / `Maximum concurrency for M tokens per
+  request: X.Xx` is the capacity number. vLLM's `Prefix cache hit rate:` line
+  is a LIFETIME average across every prompt the replica ever saw — send the
+  same prompt twice to an idle engine and compare TTFTs (measured 22.3 s cold
+  → 1.4 s warm for a 50k prefix). `GPU KV cache usage` says the pool is not
+  full; it does NOT say a prefix survived. The KV budget is NOT monotonic in
+  chunk size (5.44 GiB at CHUNK 8192, 4.22 at 16384, 4.16 at 4096) — measure
+  three points before treating a knob as a lever.
+- **Soak before believing:** 15–30 min of concurrent families, count failures
+  AND latency; report hit/miss split and medians with min/max — the median of
+  a bimodal distribution describes neither mode.
+- **Verification that is not the benchmark:** run the client the fleet
+  actually uses; and a deployed unit is not proven until an unattended reboot
+  brings it back.
+- **Record the "no" answers** with numbers and mechanism — placement balance,
+  concurrency admission and chunk-size tweaks all measured flat or worse
+  here; the wall was arithmetic (cached prefixes + live sequences share one
+  pool).
+
+## 6. Sweep hygiene (the arms that produced wrong verdicts)
+
+- **Prefetch uniformity:** every arm in a sweep carries the same
+  expert-prefetch setting. The 2026-09-29 campaign measured losing arms with
+  prefetch ON and incumbents OFF; untaxed, verdicts inverted (no-spec
+  12.11 → 41.47; gather 32×64 parity → +18.7 %). Never re-read a taxed
+  verdict.
+- **Knob-landed preflight:** for every `-e` knob, confirm
+  `podman exec <ctr> printenv <VAR>` — a knob the fork silently ignores
+  produces a "parity" arm that is really the default (v1 of the gather sweep
+  lost all 3 arms this way; serve-script hooks are additive and can drop
+  vars).
+- **Always run an explicit OFF arm:** the 2026-09-29 campaign's only win was
+  deleting a baseline's silent `--expert-prefetch 1` (+40–51 %). Challengers
+  compare against default-off, not "incumbent + one flag".
+- **Pin the capacity bar:** `--kv-cache-memory-bytes` (contract) rather than
+  autofit (remainder) so pool numbers are reproducible; compute need =
+  tokens × measured B/token (hybrid models: KV dtype does NOT double mamba
+  states — measure, don't assume 2x).
+- **Report decode AND prefill (TTFT) per rung:** a decode-only view nearly
+  deployed a 2x-prefill-tax config. Per-depth tables, never a single median —
+  one fork's gather win grows with depth (54 t/s @1k, 39.5 @short).
+- Container recreate (`podman rm -f`) wipes `podman cp`'d harnesses — re-copy
+  `mtp_measure.py`/`bench_iter.py` after EVERY recreate, or mount them.
+
+## 7. Bash laws that ate windows (each cost an arm)
+
+1. `${VAR:+SPEC=$1}` in front of a command expands to a bare word →
+   `SPEC=3: command not found`, nothing runs. Pass literal assignments:
+   `SPEC="${1:-}" bash serve.sh`; the callee's `${SPEC:+--flag}` skips empty.
+2. `${2WAY:-0}` is fatal `bad substitution` (digit-leading name), and inside
+   an if/else it aborts ONLY that compound statement — the script CONTINUES,
+   every arm tears down clean with zero measurements. Detection law: `ls`
+   the per-arm `.measure.txt` files before trusting any summary; arm echoes
+   are not evidence.
+3. `local a=$1 b=$x` expands `$x` before the builtin assigns — under `set -u`
+   it aborts AFTER the harness stopped production. Two statements.
+4. `sed -i` renames the file; already-running bash keeps the OLD inode —
+   mid-flight patches miss arms launched in the preceding minutes. Patch
+   early or run a make-up window.
+5. Never heredoc python through ssh→zsh→bash — write to a file, ship it
+   (`ssh host 'cat > /path && chmod +x /path' < local.sh`); nested quoting
+   fails silently. Same trap in `git commit -m`: backticks become command
+   substitution.
+6. No comments INSIDE a backslash-continued command (the continuation joins
+   the comment); verify multi-line command construction with `bash -n` AND a
+   render test.
+
+## 8. Results location and recovery (frank-primary arms)
+
+| copy | path |
+|---|---|
+| per-arm evidence | `$ARMS_DIR` = `~/.jobs/<run-stamp>/` (`.rc`, `.done` beside it) |
+| summary | `$ARMS_DIR/summary.tsv` = `arm rc decode_tps decode_tps_t0.7 kv_tokens boot_s` + `<arm>.out/.env/.overlay/.rc/.metrics` |
+| server-side last window | `/home/dank/.autoresearch-frank-primary/run.json` |
+| durable memo | `.../last/run.json` + `last/hash` (promoted only after bench AND verified restore) |
+| fallback | `~/.omp/agent/sessions/<ws>/<session>/<N>.{shake,bash}.log` (permanent) |
+
+`/tmp/frank-primary-arms-*` and `/tmp/armstub/` are stub self-tests —
+constants 99.9/88.8 identical across arms by construction; **identical
+numbers across arms == stub or bug, never a measurement.** After compaction,
+re-harvest with `grep -rhoE 'decode_tps[a-z0-9_.]*=[0-9.]+'` over the session
+logs. A window costs ~20 min warm; a detached runner's timeout ≥ 3600 s per
+window.
+
+## Related
+
+- [[RDNA4 MoE Expert Offload VRAM Budget]] — the plan arithmetic being swept
+- [[DFlash DSpark Speculative Decoding]] — spec A/B protocol and shapes
+- [[LLM Performance Diagnosis]] — profiling and attribution methods
+- [[Remote Job Runner]] — detached execution of long sweeps
+- [[frank-medium-llamacpp-dflash2]] — the measured ubatch-optimum history

serverAppConfigs-docs/LLM/LLM Performance Diagnosis.md

@@ -0,0 +1,221 @@
+---
+type: reference
+title: LLM Performance Diagnosis
+description: Probing why a served LLM is slow — the flat-depth probe, the expert-offload duty-cycle arithmetic, kernel-time attribution via the profiler window, live telemetry without benchmarking, RAM-bandwidth ceilings, IBS tensor attribution and prefix-cache-aware routing.
+tags: [performance, profiling, moe, pcie, prefix-cache, routing, telemetry]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# LLM Performance Diagnosis
+
+When a served endpoint is slower than wanted and someone names a suspect
+("PCIe is saturated", "the gather is the bound"), throughput arithmetic cannot
+settle it — it is usually circular (it assumes the suspect consumes the step).
+Get data in this order; the profiler window alone has retired multi-day
+optimization plans.
+
+## 1. Is decode context-bound at all? (the flat-depth probe)
+
+Sweep decode across context depth. **Flat** ⇒ attention over history is not
+the cost; the problem is per-token expert movement and KV/attention tuning is
+ruled out entirely. A genuinely context-bound server shows a clear downward
+slope. Record cold vs warm separately — a small gap means the cache is
+working. Measured on the fp8 expert-offload shape: 12.11 tok/s FLAT from 304
+to 16,095 tokens.
+
+## 2. Duty-cycle arithmetic (the "GPU not fully utilized" answer)
+
+Symptom: smooth ~25–35 % utilization with 100 % peaks on an expert-offload
+MoE. This is the EXPECTED signature, not a fault:
+
+```
+per decode step: compute burst ~100 ms (100 %) → expert-gather wait ~400 ms (~0 %)
+resident fraction = arena slots / total experts   (e.g. 6574/24576 ≈ 27 %)
+PCIe bytes/token  = miss_rate × lookups × slab_bytes  (0.73 × 480 × 2.344 MiB ≈ 0.84 GB)
+predicted duty    ≈ compute_ms / (compute_ms + wait_ms)
+```
+
+Validate the terms against the deployed engine's boot plan
+([[RDNA4 MoE Expert Offload VRAM Budget]]). If predicted ≈ observed, the low
+average is structural. Note small-transfer reality: bulk bandwidth (91 % of
+ceiling at 1 GiB) overstates what 2–3 MiB slabs get (~72 % plus ~20 %
+per-transfer latency). Before blaming the link, measure it
+([[PCIe Link and Atomics]]) and compare the engine's effective gather rate to
+the measured slab-size bandwidth — a 40–50 % gap is the software read path,
+not the wire. `rocm-smi --showuse` reads ~0 % during the whole multi-minute
+weight load: show VRAM (`--showmeminfo vram`) as the load-progress signal.
+
+## 3. Kernel-time attribution (the profiler window)
+
+Run BEFORE building any replacement kernel.
+
+1. Stop + mask production; boot the shape under test with ONLY the profiler
+   config added. vLLM argv:
+   `--profiler-config '{"profiler":"torch","torch_profiler_dir":"/prof","torch_profiler_with_stack":false,"torch_profiler_record_shapes":false,"torch_profiler_with_memory":false,"torch_profiler_dump_cuda_time_total":false}'`
+   — `with_stack=false` is load-bearing (with stacks the stop hangs and the
+   trace buffer grows ~40 GB → kernel OOM-kill of the engine core, anon-rss
+   ~95 GB; the worker-side stop can hang 5 min on ROCm/kineto). Bind-mount
+   `/prof`. If a shell substitutes values into that JSON, verify the escapes
+   actually landed.
+2. Window it: warm request → `POST /start_profile` → N generations of ≥128
+   tokens (~15 s of traffic) → `POST /stop_profile` (client may time out at
+   60 s; the trace still lands). `telemetry_report` has ONE caller: the
+   gpu_worker profiler-stop path. Overhead is small enough to keep decode
+   comparable (9.6 vs 9.9–10.0 tok/s measured).
+3. **Parse ON THE SERVER** (traces are 100–150 MB gzipped, GB expanded).
+   Rank `ph=="X"` events with `cat in {kernel, gpu_memcpy, gpu_memset}` by
+   summed `dur` (µs): report % of device time, launch count AND µs/call
+   (22 µs × 708k launches is a different problem than 4 ms × 4k), plus the
+   first-to-last event span (fraction of wall the device was busy). Aggregate
+   by FULL mangled name (template args are the tuning signal) and by stripped
+   symbol. **Trap: `gpu_user_annotation` events masquerade as GPU time** (a
+   naive sum reported 63 % in one `execute_context_0(...)` entry; spans
+   overlap and can exceed the wall clock).
+4. Divide by the window's token count → ms/token; if kernel time ≈ step
+   time, the table IS the optimization plan.
+
+Worked example (GLM-5.3-Flash, 2×GPU, offload, decode 9.9 tok/s):
+
+```
+34.8 %  708,500 launches  vectorized_elementwise_kernel   <- the actual bound
+10.0 %   42,000           MoE W4A16 GEMM
+ 8.9 %   83,160           router/top-k (sort-based)
+ 6.5 %  133,920           hyper-connection pre/post kernels
+ 1.05 %   20,664          <the long-suspected expert gather>
+```
+
+No amount of link arithmetic would have produced those two conclusions: the
+suspect was 1 %, and the mass was pointwise glue from an uncompiled trunk —
+whose own boot log names the cause (`FULL_DECODE_ONLY` override). It also
+predicted a negative result in advance (spec decode cannot help a shape whose
+per-token cost scales with the token count) which matched an unexplained flat
+MTP measurement.
+
+## 4. Telemetry without benchmarking (live hosts)
+
+When benching is forbidden (production serving), mine the runtime's own
+per-request telemetry — it measures the real workload and costs nothing.
+
+- llama.cpp (root unit, `-lv 3`): `sudo journalctl -u <unit> --since "-3h"
+  --no-pager` → `slot print_timing` lines (prompt eval, eval, total, `graphs
+  reused`, `draft acceptance … mean len`, `n_gen = N, tg = X t/s, tg_3s = …`).
+- halogen (rootless quadlet): `journalctl --user -u <unit>` (**never sudo** —
+  per-user storage) → `serve_api: mtp 28 tok in 0.66s = 42.45 t/s | 16
+  rounds, commit 1.75/round | prompt 58, prefill 0.92s | detok 24us/tok`.
+- Read the envelope, never one sample: short-prompt `prompt eval` is
+  overhead-dominated (57 tok @ 168 t/s) — quote the longest available (1106
+  tok @ 1013 t/s); decode and acceptance are content-dependent (0.46–0.76
+  measured on one unchanged build); `graphs reused > 0` / `selected slot by
+  LCP similarity` prove prefix-cache reuse.
+- **Attribute PCIe without touching the model:** background `nvidia-smi dmon
+  -s t -c 16 -o T`, ONE ordinary `max_tokens=200` request, `wait` at the end
+  (without it ssh closes and the client dies — the journal shows `W srv stop:
+  cancel task` and NO print_timing while curl still reports plausible
+  status). Measured: peak ~230 MB/s against a ~7,900 MB/s link = ~3 %, 0 at
+  idle — the sampled-logits stream, not the transport, is decode traffic on
+  that shape.
+- One `max_tokens=200` request is production traffic and fine; anything that
+  pins the card is a bench → stop-and-mask first. Check WHICH instance you
+  are reading (`systemctl status | head -12`; task ids jump across
+  restarts). The engine's own ledger (`KV self size`, `n_slots`,
+  `n_ctx_slot`) outranks your model of it.
+
+## 5. The fork's data path (before blaming the link)
+
+Offload engines pick between DMA copy-in and read-through UVA dereference
+inside the GEMM. Grep the installed package for the LRU/offload module and
+read the BRANCH CONDITION, not just the env names. A token-count threshold
+decides the path, and it interacts with spec decoding: an MTP step is k+1
+tokens, so the step can silently cross the threshold onto the slow path as
+draft depth grows. Check the threshold's premise: a default tuned at
+12.4 MiB/expert does not transfer to 2.344 MiB/expert, where copy-in can be
+several times faster. TunableOp ghost-skip: `CLAV_TUNABLEOP_SWEEP=1` alone
+does NOTHING under expert offload (`CLAV_TUNABLEOP_WITH_OFFLOAD=1` required);
+check for the skip line before concluding autotuning ran. Gather-geometry
+knobs read at import time → each arm is a full reload.
+
+## 6. RAM-bandwidth-bound MoE ceiling (before recommending a quant)
+
+```
+routed experts/token = top_k × n_layers
+bytes/token          = routed_params × bytes_per_param + shared_expert_bytes
+ceiling (tok/s)      = measured_bandwidth / bytes_per_token
+```
+
+Measure the HOST's real bandwidth, never assume DIMM ratings: 4-DIMM AM5
+boards downclock to DDR5-3600 regardless of 5600 sticks (two large DIMMs beat
+four small). Single-threaded reads under-report dual-channel ~4x — use the
+multi-threaded plateau of a checksumming reader (`scripts` pattern:
+`/tmp/bw.c`, 1/2/4/8/16 threads). Worked example (DeepSeek-V4.1-Flash):
+6×40 experts × 35.4M params = 8.5B params/token; at 3 bpw + 0.53 GB shared =
+**3.7 GB/token** → 42 GB/s ⇒ **11.3 tok/s ceiling**. Sanity anchors: H100
+NVL + CPU offload 226 GB ⇒ 5.8–9.0 tok/s; 2×3090 + 768 GB DDR4 ⇒ 12.2.
+CPU compute rarely binds below 4 bpw (~5.3 flops/byte ⇒ ~223 GFLOPS at 42
+GB/s); PCIe expert STREAMING is worse than CPU compute (Gen5 x8 ~22.7 GB/s
+effective < half of RAM); prefill reads the union of a batch's experts
+(prefill ≈ bandwidth / total_expert_bytes × ubatch). A quant that fits only
+by ignoring the Engram table does not fit (Engram = sparse n-gram lookup,
+NVMe-tolerant; the routed experts are not). An iGPU GTT aperture is an ALIAS
+of the same RAM, not extra memory.
+
+## 7. IBS model-access attribution (AMD CPUs)
+
+Real CPU access patterns for mmap'd weights, live, no stop (established on
+frank-primary, 2026-09-26, ~2.7M attributed samples in 75 s). Reference:
+`scripts/frank-primary-expert-activity.py` (self-contained dump + sample +
+analyze; `--analyze-only`).
+
+1. GGUF header-only scan (ms for a 50 GiB part) → tensor extents. **Extents
+   from CONSECUTIVE offsets, never a type-size table** — a wrong per-element
+   size silently under-covers every tensor and dumps tail samples into a junk
+   bucket (first run lost 59 % of expert samples; a uniform gap/exps ratio
+   ~1.4 across parts is the tell). Data start = `(tell + 31) & ~31`.
+2. `/proc/<pid>/maps` VMAs (7-group regex; pgoff = group 4, path = group 7)
+   matched against part files; multi-VMA per part is normal.
+3. `sudo perf record -e ibs_op// -c 50000 --data -o out.perf -p <pid> --
+   sleep <dur>` WHILE a mixed workload runs (idle perf numbers mean nothing).
+   `sudo perf script -i out.perf -F addr` for addresses.
+4. Attribute: bisect VMAs, `foff = pgoff + (addr − vma_start)`, bisect tensor
+   starts; per-expert index = `int(lo / (span / N_EXPERTS))` (expert-major
+   rows).
+5. Verify before believing histograms: gap ≈ 0, every tensor-bearing part
+   nonzero, and MoE blocks should be FLAT by construction (top-k fires every
+   block every token) — a non-flat block histogram means mis-attribution
+   first. Established law: expert routing is per-block flat, per-EXPERT
+   skewed.
+
+## 8. Prefix-cache-aware routing (multi-replica serving)
+
+Measure with a **sequential-turn client**, not only a batch benchmark — a
+batch of distinct prompts parallelises prefill across replicas and hides the
+problem. The two failure modes: round-robin loses every other cache hit
+(90k-token contexts: warm-turn TTFT 2676 ms round-robin vs 1096 ms pinned);
+pinning-everything loses a card (16 requests: 18.6 s spread vs 30.1 s
+pinned). **The rule that gets both: route by request shape** — a request
+carrying more than one message is a continuation (agent loop re-sending its
+context) → pin to the replica that served that prefix; anything else → spread
+with a COUNTER, not a prefix hash. Add failover: demote a replica on
+connection failure and retry on a live one BEFORE writing anything to the
+client (safe pre-stream; never retry after streaming started).
+
+The wall no policy fixes: **a running sequence and its cached prefix draw on
+the SAME pool.** Three concurrent 50–80k families against 122,636 tokens/card
+thrash under every placement and admission policy tried (counter placement,
+load-balanced placement, one-large-request-per-card admission, both chunk
+directions — the last COSTS KV in both directions). When the working set
+exceeds aggregate KV, say so: the levers are capacity, fewer concurrent
+large contexts, or a different platform. Traps: a lifetime hit-rate counter
+is not a per-request verdict (probe directly: same prompt twice to an idle
+replica — 22.3 s cold → 1.4 s warm); salt corpora; record sample sizes;
+`vllm bench serve` needs `--tokenizer <local path>`; verify metric names
+against the live endpoint before dashboards. Test shape: drive the routing
+function directly with synthetic bodies (stdlib-only policy module);
+conversation identity is the FIRST N KiB of prompt — a shorter tested prefix
+hashes differently per turn and pinning silently never happens.
+
+## Related
+
+- [[LLM Bench Windows and Measurement]] — the window discipline and usage-counter law
+- [[PCIe Link and Atomics]] — link probing and the streaming-vs-compute ceiling
+- [[RDNA4 MoE Expert Offload VRAM Budget]] — the plan behind duty-cycle terms
+- [[vLLM Porting and Compile Triage]] — the compile fix for the pointwise-glue bound

serverAppConfigs-docs/LLM/Llama Cpp Multi-Device Splits.md

@@ -0,0 +1,274 @@
+---
+type: reference
+title: Llama Cpp Multi-Device Splits
+description: What llama.cpp can and cannot express across devices, the no-collective layer-split path, split balancing by measurement, drafter KV sizing from GGUF headers, and heterogeneous placement triage.
+tags: [llama.cpp, multi-gpu, layer-split, pcie, sizing, speculative]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Llama Cpp Multi-Device Splits
+
+## 1. Expressibility law — classify the arrangement before running anything
+
+llama.cpp expresses exactly two things across devices, and they are not
+interchangeable:
+
+| arrangement | how | expressible? |
+|---|---|---|
+| **Layer split** — layers 0..n on dev A, rest on dev B | `-sm layer -ts <a>/<b>` | **yes**, no collectives needed; works cross-vendor (CUDA + Vulkan measured 2026-09-16) |
+| **Per-tensor placement** — attention+KV on A, experts on B | `-ot <regex>=<dev>` | **NO.** `--override-tensor` sets `has_tensor_overrides()`, a REQUIRED term of the scheduler's `pipeline_parallel` flag (only read site: `llama-context.cpp`, in the `pipeline_parallel = …` expression). That flag sets `n_copies = parallel ? MAX : 1` in `ggml_backend_sched_new`; with `n_copies == 1` a pre-allocated weight on a non-primary device has no backend and the graph aborts at `ggml-backend.cpp:941` (`pre-allocated tensor (blk.0.ffn_down_exps.weight) in a buffer (Vulkan1) that cannot run the operation (NONE)` — `(NONE)` is the leaf's own op name). |
+
+Failure modes observed (mainline `37b53fd`): per-tensor `-ot exps=Vulkan1`
+with `-dev CUDA0,Vulkan1` → scheduler abort; layer-scoped `-ot` with a
+cross-backend pair → model load fails with only `failed to load model`;
+`-ot x=CPU` **works** because the CPU backend is last in the list — **the CPU
+is the only legitimate partner in a heterogeneous split.** On a CUDA+AMD host
+a card can be the trunk OR the offload host, never the complement of an
+AMD-resident peer. Pure Vulkan changes nothing here — the gate is the
+override, not the driver mix.
+
+If the plan needs tensors on a specific device: stop, the answer is no for
+any vendor mix. Expressible alternatives, in order: (1) two independent
+endpoints, one model per device; (2) single-vendor layer split; (3) device +
+CPU per-tensor offload (mainline pinning guard: `GGML_CUDA_NO_PINNED`; ik fork
+was `GGML_CUDA_NO_PINNED_WEIGHTS` — without it a CPU-overridden tensor can
+land in a pinned `CUDA_Host` buffer and page-lock ~108 GiB).
+
+Layer-split arithmetic to check before promising: a whole layer carries its
+experts (MoE) — compute GiB/layer, not per param-group; only some layers carry
+KV (every 4th on the hybrid), so a share of layers is only a share of KV; big
+non-repeating tensors (PLE tables) land on the main device and cannot move —
+they cap how small the primary can get.
+
+## 2. No collectives? The atomics wall and the layer-split answer
+
+Decide with one line before designing:
+
+```bash
+sudo lspci -vv -s <card2> | grep AtomicOpsCtl   # ReqEn- ⇒ no collectives, ever
+lspci -tv                                        # a Routing- bridge is a hardware wall
+```
+
+`ReqEn-` kills vLLM entirely — TP needs a per-layer allreduce AND PP needs a
+device-backend `send`/`recv` of the hidden state
+(`vllm/distributed/parallel_state.py`) — i.e. the same RCCL primitive that
+fails on an atomics-less card. llama.cpp's `--split-mode layer` needs no
+collective: layers distribute in order, only the boundary activation crosses
+PCIe once per forward pass. Full probe/repair detail:
+[[PCIe Link and Atomics]].
+
+## 3. Build: pin by feature, bundle the runtime at the exact depth
+
+- Pin the ref by FEATURE list (e.g. DFlash2 + qwen35 MTP on 2026-09-16 →
+  `ggml-org/llama.cpp#27342`); never float. Verify a fork build the same way
+  (`ls <venv>/…/spec_decode/ # dflash/ vs dflash2/`).
+- `-DCMAKE_HIP_ARCHITECTURES=<gfx only>` — a sibling arch links, loads, then
+  segfaults. `-DGGML_HIP_RCCL=OFF` when there is no atomics path (only tensor
+  mode degrades to the butterfly fallback; layer split unaffected).
+- **TRAP — `cp -r SRC DST/` flattens when `DST` does not exist** and drops one
+  directory level. rocBLAS/hipBLASLt look for device code at a FIXED depth
+  (`../lib/rocblas/library`, `../lib/hipblaslt/library/<arch>`); flattened,
+  the engine loads every GiB, answers `/health`, and dies on the FIRST request
+  with `Cannot read .../TensileLibrary.dat: No such file or directory for GPU
+  arch`. `mkdir -p DST && cp -r SRC/. DST/`, then
+  `patchelf --set-rpath '$ORIGIN/../lib' build/bin/llama-*`.
+- Dual-backend build (CUDA + Vulkan is the practical pair; HIP+CUDA do not
+  coexist): `-DGGML_CUDA=ON -DGGML_VULKAN=ON` in one tree; Vulkan needs
+  `glslc` + libvulkan headers. **Device names are not what you expect:**
+  `--list-devices` lists the discrete card under BOTH backends (`CUDA0` AND
+  `Vulkan0`) and the APU LAST (`Vulkan1`) — a wrong `-ot` name makes
+  llama.cpp print the valid set and exit (self-diagnosing).
+- What the existing build can do: `GPU_BACKEND_*=OFF` lines in its
+  CMakeCache.
+
+## 4. Serve flags that matter on a split
+
+```bash
+HIP_VISIBLE_DEVICES=0,1 llama-server \
+  --device ROCm0,ROCm1 --split-mode layer --tensor-split 1,1 \
+  --gpu-layers 999 --fit off --flash-attn on \
+  --ctx-size <n> --parallel 1 --cache-type-k f16 --cache-type-v f16 \
+  --jinja --metrics --cache-prompt \
+  --spec-type draft-dflash --spec-draft-model <dflash2.gguf> \
+  --spec-draft-device ROCm1 --spec-draft-n-max 7
+```
+
+- `--fit off` + explicit `--ctx-size`: auto-fit must not resize a context set
+  on purpose. **`--parallel N` DIVIDES `--ctx-size`** — one slot at large ctx
+  is the concurrency policy, concurrent agent loops queue.
+- Pin the drafter to the card holding the target's LAST layers
+  (`--spec-draft-device`); under tensor mode an unpinned draft joins the
+  meta-device KV allocation and aborts the KV build. `--spec-draft-n-max` must
+  be ≤ the drafter's `block_size - 1` (DFlash2: 8 ⇒ 7).
+- Qwen3.8-family templates RAISE on `reasoning_effort=high` (500) — probe
+  low/medium/high/xhigh and map harness levels before declaring ready.
+
+## 5. Split balancing: measure, never interpolate
+
+**A `-ts` step is ONE LAYER, and hybrid stacks do not have uniform layers.**
+qwen35 is 16 × (3 Gated DeltaNet + 1 Gated Attention): one step ≈ 0.43 GiB on
+a DeltaNet boundary, ≈ 1.43 GiB on a full-attention one (weights + that
+layer's own f16 KV) — measured on frank-medium. The gap can jump 2.9 GiB in a
+single step; the best integer split can sit 0.2 GiB from balance while its
+neighbours are 1+ GiB off. An affine fit reproduces the deployed point and
+overstates gaps in between (`tests/frank_medium_unit.bats` encodes the fit as
+a gate; treat a NEGATIVE fitted constant as a unit bug).
+
+Knobs that move weight and REQUIRE a re-measure: `--ubatch-size` (compute
+buffers are **~2x heavier on the output/drafter card** — 512→1536 added
+~3.3 GiB there vs ~1.7 GiB elsewhere), a drafter swap (weights + draft KV),
+`--ctx-size`/`--parallel` (KV). Total free VRAM is invariant across splits —
+only the distribution moves; use that to sanity-check a suspicious row.
+
+**Procedure (one split per load):**
+
+```bash
+WORK=/tmp/split.env; cp scripts/serve-<svc>.env $WORK
+export FRANK_MEDIUM_ENV_FILE=$WORK
+for ts in 39,26 38,27 37,28 36,29 35,30; do
+  sed -i "s|^TENSOR_SPLIT=.*|TENSOR_SPLIT=$ts|" $WORK
+  nohup bash scripts/serve-<svc>.sh >/tmp/split.log 2>&1 &
+  # poll /health (or process exit), then:
+  rocm-smi --showmeminfo vram | grep 'VRAM Total Used'   # sample ~6 s after health
+  pkill -x llama-server; sleep 6
+done
+```
+
+Load → sample → kill, NO requests sent (allocation only). Run under `nohup`
+and poll from a separate call (a load is 30–90 s). Pick the smallest
+inter-card gap that clears the per-card floor with room to spare; one `-ts`
+step ≈ 1.5 GiB at ~6k batch / 262k ctx on 30 GiB cards — enough to bracket,
+not to solve. Prefer balance: the busy card's headroom bounds the context
+(`ctx_extra ≈ (floor − measured_free) / per-position cost on the busier
+card`). A faulted split (`SDMA0 UTCL2 page fault` → `illegal memory access` →
+`ggml_abort`) is OFF-LIMITS — re-measure, never re-tune; more headroom does
+not prevent a boundary-copy fault.
+
+## 6. Drafter KV: sized from the TARGET, bounded by its window policy
+
+`common/speculative.cpp` builds the draft context with
+`cparams.n_ctx = llama_n_ctx(ctx_tgt)` — **no flag caps it.** So:
+
+```
+draft KV = ctx_tgt × block_count × 2 (K,V) × kv_heads × key_length × dtype
+```
+
+…unless the drafter declares a sliding window. Measured at 262,144 target
+ctx: DFlash2 (`sliding_window = 2048`) ≈ 40 MiB; a same-shape 5-layer drafter
+with NO window = 20 KiB/position = **5.0 GiB** — a four-layer swing in the
+balance point, and swapping them invalidated an already-balanced split.
+
+**Failure signature:** the engine loads the whole model, answers `/health`,
+and dies on the FIRST draft decode with `ROCm error: out of memory` /
+`cudaMalloc failed` inside `common_speculative_impl_draft_dflash::process` —
+a sizing failure, not a broken drafter; under a self-healing unit it presents
+as a crash-loop. Read `sliding_window`, `block_count`, `head_count_kv`,
+`key_length` from the drafter's GGUF header first. A compact reader (the
+three format traps: no KV alignment — v3 alignment applies only to the tensor
+section; arrays are `type(9) → element_type(u32) → count(u64)`, element type
+FIRST; the enum is `7=BOOL 9=ARRAY`, 64-bit = 10/11/12, and 8=STRING/9=ARRAY
+is the v2/v3 file order, NOT gguf.h's 2/3):
+
+```python
+import struct, sys
+def rd(path):
+    f = open(path, 'rb'); assert f.read(4) == b'GGUF'
+    struct.unpack('<I', f.read(4)); struct.unpack('<Q', f.read(8))
+    nk = struct.unpack('<Q', f.read(8))[0]
+    def s():
+        n = struct.unpack('<Q', f.read(8))[0]; return f.read(n).decode('utf-8', 'replace')
+    SCALAR = {0:1,1:1,2:2,3:2,4:4,5:4,6:4,7:1,10:8,11:8,12:8}
+    FMT = {0:'<B',1:'<b',2:'<H',3:'<h',4:'<I',5:'<i',6:'<f',7:'<?',10:'<Q',11:'<q',12:'<d'}
+    keys = {}
+    for _ in range(nk):
+        k = s(); t = struct.unpack('<I', f.read(4))[0]
+        if t == 8: keys[k] = s()
+        elif t in SCALAR: keys[k] = struct.unpack(FMT[t], f.read(SCALAR[t]))[0]
+        else:
+            et = struct.unpack('<I', f.read(4))[0]; n = struct.unpack('<Q', f.read(8))[0]
+            if et == 8: keys[k] = [s() for _ in range(n)]
+            elif et in SCALAR: keys[k] = ('arr', et, n); f.read(n * SCALAR[et])
+            else: raise Exception('unsupported arr elem type %d' % et)
+    return keys
+```
+
+## 7. Heterogeneous placement triage (go/no-go with arithmetic)
+
+- **Link:** `LnkCap` vs `LnkSta` at every hop; expect **idle-downshift** (a
+  Gen4 x4 dock idles at Gen1 x4, ~1 GB/s) — measure under load. Per-hop cost
+  vs hidden state (5120-dim f16 = 10 KiB/hop ≈ 1.3 µs at Gen4 x4): bytes are
+  almost never the constraint; **hop count is** — every boundary is a host
+  sync, and a per-op-split runtime can hit 100 splits/token at ~5 % SM. Read
+  `graph splits` from the loader log.
+- **KV is usually the bottleneck:** `KV/token = kv_bytes_per_position ×
+  context` (attention reads the whole cache every token). Highest-leverage
+  placement: **KV on the fastest-memory device together with its attention
+  compute** — never KV alone over the link (GBs over 7.88 GB/s = hundreds of
+  ms/token). Check fit: `kv_positions × slots × bytes + attention_trunk ≤
+  VRAM`.
+- **What placement never buys:** prefill is compute-bound and expert weights
+  are read once per batch — moving them buys nothing and adds a serial stage.
+  Calibration: Strix Halo + 3090, moving 12.5 % of expert mass to the GPU
+  gave decode +12 % but prefill 192 → 32 tok/s. The harmonic
+  `1/(f_big/BW_slow + f_fast/BW_fast)` reproduces the decode gain within ~2
+  points (assumes sequential stages — true at batch 1).
+- **UMA check before assuming additivity:** `mem_info_vram_total` (BIOS
+  carve-out) vs `mem_info_gtt_total` (aperture onto SYSTEM RAM). A big GTT
+  window means the "GPU memory" IS the same LPDDR: adding a stage there adds
+  a compute engine, not capacity/bandwidth. Only discrete VRAM is additive.
+- **Speculation as the "cheap" split:** bandwidth-free (token ids per round)
+  but needs the exact tokenizer for a plain drafter or a head trained for
+  that target; a plain drafter costs K sequential forwards (break-even needs
+  p ≈ 0.7+ at r≈0.4–0.5, K=3); compare against the INCUMBENT (a target
+  already committing ~1.7 tok/round on its own MTP head is a high bar); and
+  check the runtime accepts external drafters at all (closed engines 400
+  unknown drafter names by design). Details: [[DFlash DSpark Speculative Decoding]].
+- **Hazards:** keep paged/sidecar tables where they are read from; keep mmap
+  ON for huge-embedding models (anonymous memory → cgroup kill); CUDA graph
+  capture can abort on unusual graph shapes; make the FIRST measurement a
+  single-device baseline of the runtime on the new device — a split's
+  arithmetic is worthless if the baseline loses to the incumbent; a split
+  consumes the fast device entirely — check what it was serving.
+
+**Verdict ranking:** (1) two independent endpoints — often the real answer;
+(2) one process, weights-resident, KV co-located with attention on the fast
+pool; (3) hot-expert cache only if `misses × bytes × tok/s` fits the link and
+overlaps — no mainstream runtime implements it; (4) name what does NOT help:
+TP across unequal devices, KV on the far device, loading over the link,
+prefill-motivated placements.
+
+## 8. Harness syntax and verification
+
+`llama-bench` separates a LIST with `/` and SWEEP values with `,`:
+`-dev CUDA0,Vulkan1` is TWO single-device runs, never one two-device run;
+`-ts 0.85,0.15` is two sweep values each becoming `[0.85, 0, …]` — device 0
+takes everything. Write `-dev A/B -ts x/y`. Read the table's own `dev` column;
+any placement-dependent number needs an independent memory check
+(`nvidia-smi` sampling while loaded). (This artifact produced a wrong fleet
+conclusion 2026-09-16 — "CUDA + Vulkan cannot serve one model" — with two
+bogus OOMs that were really all-111-GiB-on-the-24-GiB-card.)
+
+Verification battery: `/health`; both cards' memory ≈ half the weights each
+(busier card also carries the drafter; compare `model_bytes/2 + drafter +
+ctx×KV/2 + buffers`); raise ctx by N and watch BOTH cards rise by
+N·bytes/2 — neither card's total may equal the model file (that arithmetic
+rules out replication); `spec_decode_num_drafts_total` > 0 and a real
+completion; a deep prompt (140k) proves a raised window is REAL, not
+declared; repeat prompt → `timings.prompt_n` collapses (prefix cache live);
+`/slots` shows expected `(id, n_ctx)` pairs; two concurrent requests with
+`journalctl -o cat | grep 'slot print_timing'` proves drafting on BOTH slots
+(default journal output truncates these lines).
+
+Ops traps: a server started over ssh outlives stopping the runner (keeps the
+port — `pkill -x <binary>` first); `systemctl start` of a root unit needs
+sudo (polkit "Access denied" otherwise); old crash `ROCm error` lines live in
+the journal forever — check `ActiveEnterTimestamp`/`NRestarts` before
+treating a grep count as a live fault.
+
+## Related
+
+- [[PCIe Link and Atomics]] — the atomics wall, link probing, repair
+- [[Llama Cpp Serving Sizing]] — slot-pool and KV pricing
+- [[DFlash DSpark Speculative Decoding]] — spec types, chaining, A/B
+- [[frank-medium-llamacpp-dflash2]] — the layer-split era runbook
+- [[Dual-R9700-Cooperation]] / [[x4-pair-experiment]] — the measured history

serverAppConfigs-docs/LLM/Llama Cpp Serving Sizing.md

@@ -0,0 +1,163 @@
+---
+type: reference
+title: Llama Cpp Serving Sizing
+description: Slot-pool sizing against a headroom band, per-device VRAM accounting, prompt-cache/state sizing and the QSA flash-attn fix — the measurement-first procedures for llama.cpp serving shapes.
+tags: [llama.cpp, vram, kv-cache, slots, flash-attn, sizing]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Llama Cpp Serving Sizing
+
+## 1. Slot-pool sizing against a headroom band
+
+Goal: choose `CTX_SIZE`/`PARALLEL`/KV types so N conversations fit a GPU while
+keeping a requested amount of VRAM free — from measurement, never arithmetic
+guesswork.
+
+1. **Fix the invariants.** Read the payload's KV geometry (attention layers
+   only, KV heads, head_dim, KV dtype bytes) and the per-slot rule
+   `n_ctx_slot = CTX_SIZE / PARALLEL`. A hybrid model with
+   `kv_unified = 'false'` gives each slot its own cache AND a fixed per-slot
+   state buffer — slot count is a memory term, not free.
+2. **Two boots on the real card** (production stopped, per the bench law),
+   each near the target; record `nvidia-smi` used/total plus the boot log's
+   `n_slots`/`n_ctx_slot`: shape A = current/cheap, shape B = new demand.
+   `non-KV = used − (tokens × KiB/token)`; the two rows separate
+   weights/buffers/graphs from the new demand (a 918 MB projector bills
+   ~1.4 GiB).
+3. **Solve the band:** `KV_tokens ≤ (card − non-KV − headroom_low) /
+   KiB/token`, round the pool DOWN to a multiple of `PARALLEL × 4096` so every
+   slot gets an equal integer window. Boot and measure; if outside the band,
+   walk one step — never re-derive from a formula.
+4. **Propagate the per-slot number — it is the client contract:**
+   `models.yml` `contextWindow` = `n_ctx_slot` (NOT the model's native window),
+   or long conversations die with `400 exceed_context_size_error` instead of
+   compacting; compaction/summary budgets sized to the old window get
+   re-checked; endpoint docs and the rollback table move too.
+5. **Prove behaviour:** boot-log slot count, `nvidia-smi` free in band, N
+   concurrent requests answered in parallel with N+1 queued, and — if a
+   projector ships — a solid-color image probe through an **`image_url`
+   content part** expecting the exact color. The legacy top-level `images`
+   field is silently ignored, so a probe that uses it measures a
+   hallucination, not vision.
+
+Gotchas: shorter slots can make decode FASTER (~5 % measured); the KV pool is
+allocated whole at boot so post-boot `memory.used` IS the ceiling; record both
+measured rows in the service `.env` comment block and mirror the non-KV price
+in the bats budget gate so a shape that busts the band fails the suite instead
+of the boot.
+
+## 2. Per-device VRAM accounting ("where did the memory go?")
+
+First decide which kind of memory is in question:
+
+| device | kind | read it with |
+|---|---|---|
+| RTX 3090 (Dankclaw) | discrete GDDR6X | `nvidia-smi --query-gpu=memory.total,memory.used,memory.free` |
+| gfx1151 iGPU (Dankclaw) | **UMA** — "VRAM" is host RAM | `/sys/class/drm/card1/device/mem_info_gtt_{total,used}` (114 GiB) vs `mem_info_vram_*` (2 GiB carve-out) |
+| R9700s (Dankserver) | discrete | `rocm-smi` figure (64 GiB, SVM-backed) |
+
+Per-process truth on NVIDIA:
+`nvidia-smi --query-compute-apps=pid,process_name,used_memory`. Never add the
+iGPU's GTT to host RAM as a separate pool; `free -g` is the arbiter for
+co-tenants.
+
+- **Engine ledger first.** Halogen prints weights pinned (+ remaining 2 MiB
+  contiguous blocks — the honest fragmentation signal: 240 blocks/12.3 GiB
+  after reboot vs 475/14.7 GiB before means the model did not come back into
+  the same hole), device budget, KV pool/working split, "left for everything
+  else". llama.cpp prints at `llama-model.cpp:1827` (per-backend model buffer)
+  and `llama-context.cpp:692` (compute buffer) — but **`llama-bench`
+  suppresses loader INFO and a systemd unit did not put them in the journal**
+  (verified `grep -c` = 0). Recipe: one launch through `llama-cli`/`llama-server`
+  keeping the first ~200 lines.
+- **Otherwise compose from tensor sizes.** Group the GGUF tensor table by
+  role (Flash-Next example: 75.0 GiB experts at 1.5625 GiB/layer, 32.8 GiB PLE
+  table, 3.2 GiB everything else), KV ≈ 2 KiB/position per f16 attention
+  layer, × derived context. Vendor delta is real: same layers+KV cost 12.80
+  GiB under Vulkan vs 14.99 GiB under CUDA (~3.3 vs ~1.1 GiB of
+  graph/compute pool).
+- **Traps:** mmap'd PLE is reclaimable page cache; `-ot` makes it pinned —
+  "move layers to the card" converts reclaimable cache into device memory and
+  frees nothing for co-tenants. A served model's footprint ≠ weight size
+  (frank-small: 22.6 of 24.1 GiB for an 11.29 GiB model). Do the capacity
+  subtraction before designing: 12 attn layers × 1.62 GiB + 200k f16 KV =
+  28.6 GiB > 24.1 GiB usable → KV-in-VRAM on that pair only reaches ~64k.
+- **Context lengths: derive, never remember.** `llama-bench` computes
+  `n_ctx = n_prompt + n_gen + n_depth` (defaults 512/128/0, type_k/v F16,
+  flash_attn AUTO, split_mode LAYER). `-p 512,16384 -n 64` = contexts
+  576/16,448; a plain `tg` row has ~zero depth — `-d` is the only thing that
+  puts KV in the picture.
+- **Label every number** measured / derived / inference, and state the context
+  each figure was taken at — 12.8 GiB at 65,568 tokens is not the same claim
+  as at 200k.
+
+## 3. Prompt cache and state sizing
+
+`--cache-ram` (MiB, default 8192) is HOST RAM for saved prompt states —
+unpinned, on demand, moves no VRAM. `--ctx-checkpoints` (default 32) is a
+different mechanism: per-slot KV snapshots; `0` disables them.
+
+**State law** (compute before trusting any cap): a state is the serialized KV
+of the whole sequence + constant recurrent state + the draft context's KV:
+`per_token = 2 × n_full_attn_layers × n_kv_heads × head_dim × bytes (K and V) +
+draft_per_token`. Measured on frank-medium (qwen3.5 hybrid, f16, MTP draft):
+16×4×256×2×2 = 64 KiB/token + 4 KiB draft = **68 KiB/token**; a full
+196,608-token slot state ≈ 12.8 GiB. Cross-check against the journal's
+`prompt state size` lines — the two must agree.
+
+**The journal is the instrument:** `exceeds cache size limit … skipping` =
+cap binding in production; `making room … removing oldest` / `cache size
+limit reached` = eviction, and the printed size proves the active cap. Do NOT
+claim a raise is proven by `prompt_tokens_cached_total` rising — on exact
+repeats that counter is 0 by construction.
+
+**Hybrid + zero-checkpoint trap:** for a hybrid (DeltaNet/SSM + attention)
+model, `seq_pos_min` sits at the END of the slot's own prompt (a recurrent
+cell stores only the last position), so with `--ctx-checkpoints 0` there is no
+checkpoint to fall back on and the server logs `forcing full prompt
+re-processing … (likely due to SWA or hybrid/recurrent memory)`. Consequences:
+intra-slot reuse only survives when the request echoes the assistant's
+previous reply verbatim (a pure user resend is always a full re-prefill even
+at f_sim 1.000); the RAM prompt cache becomes the ONLY carrier of a
+conversation state ACROSS slots, so a cap that refuses deep states costs
+minutes of re-prefill whenever slots are reassigned; state restoration only
+runs via the LRU path (or LCP with f_keep < 0.5) — a plain prefix hit never
+calls `prompt_load`.
+
+**Verify a change:** live argv carries the flag (not the deploy log); `/health`
+200 + real completion; VRAM unchanged and host RAM mostly free; no new
+`exceeds cache size limit` lines, with a `making room` entry larger than the
+old cap as direct proof. If the ceiling itself was not exercised live, say so —
+crossing it needs a >120k-token conversation and must not run beside a live
+service (bench law).
+
+## 4. QSA flash-attn assert fix (qwen4exp)
+
+`--flash-attn on` aborts graph build for qwen4exp models:
+`GGML_ASSERT(a->type == GGML_TYPE_F32)` at `ggml.c:5371` through
+`build_attn_qsa`. Root cause: `build_attn_qsa` calls `ggml_fill` on the
+F16-cast kq mask returned by `get_kq_mask()` when flash attn is on, but
+`ggml_fill` requires F32.
+
+Fix (mirrors the MLA sparse path in `llm_graph_context::build_attn`): run the
+top-k mask surgery on the F32 base mask (`inp->self_kq_mask`), cast to F16
+only at the attention call when `flash_attn` is enabled
+(`kq_mask_attn = cparams.flash_attn ? ggml_cast(ctx0, kq_mask_top_k,
+GGML_TYPE_F16) : kq_mask_top_k`). The patch lives at
+`scripts/patches/qwen4exp-qsa-f32-mask.patch`.
+
+Only the charlie12345/ROCmFPX build-qwen38 (PR #27742 qwen4exp merged) takes
+it; CIRU, upstream and stock builds do not. Validate live: load with
+`--flash-attn on`, health 200, a real completion finishing `"stop"`, journal
+clean of `GGML_ASSERT`/`cudaMalloc`/`missing tensor`, and flash-attn engaged
+in the timings. Context cap for full-model residency on that shape stayed
+32,768.
+
+## Related
+
+- [[GGUF Compatibility]] — what a payload is before it is priced
+- [[frank-small-llama-mimo-q6k-2x200704]] — the current two-boot priced shape
+- [[frank-medium-llamacpp-dflash2]] — where the 68 KiB/token datum was measured
+- [[DFlash DSpark Speculative Decoding]] — the spec A/B protocol on llama.cpp
+- [[Dual-R9700-Cooperation]] / [[x4-pair-experiment]] — the multi-device side

serverAppConfigs-docs/LLM/Model Swap and Rollback.md

@@ -0,0 +1,281 @@
+---
+type: reference
+title: Model Swap and Rollback
+description: The full model-swap discipline for the LLM fleet — fit preflight, fetch and payload verification, runtime build/flag checks, git-only config, live validation, systemd handoff, close-out, the three-pin revert law, and research-verification discipline.
+tags: [models, swap, gguf, huggingface, rollback, deployment, verification]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Model Swap and Rollback
+
+Applies to every endpoint swap on the fleet (frank-small / frank-large /
+frank-primary). Repo law: all config flows through git; the live models are
+host state under `/home/dank/models/<name>/`.
+
+## 1. Fit preflight — never deploy a swap without confirming it FITS
+
+1. Get the checkpoint's on-disk/weight size AND precision from the HF card
+   (`Precision: BF16` = 2 B/param; FP8/Q8 ≈ 1 B; Q4 ≈ 0.5 B). 35B params:
+   BF16 ≈ 66–70 GB, FP8/Q8 ≈ 33–35 GB, Q4 ≈ 17–20 GB.
+2. Compare against the HOST'S usable RAM (not just VRAM) for runtimes that
+   stream weights from RAM (host-RAM expert offload, llama.cpp CPU offload,
+   layer splits). Budget weights + KV pool + OS + page cache; **if weights
+   alone exceed RAM, ABORT** — pool tuning shrinks KV, it does not shrink
+   weights.
+3. **The failure mode is a self-sustaining thrash, not a crash:** a large
+   swap makes the kernel PAGE instead of OOM-killing → sshd/user space
+   starved indefinitely, no recovery window opens by itself; systemd
+   `Restart=always` turns it into a crash-reload-rethrash loop; `enabled` +
+   `WantedBy=multi-user.target` re-thrashes on every reboot. A power cycle
+   does NOT fix it.
+
+Recovery (may need console / SysRq-to-GRUB):
+
+```bash
+sudo systemctl mask --now <svc>                       # break the reboot trap FIRST
+cd <checkout> && git fetch origin && git reset --hard origin/main
+grep MODEL_DIR <env>                                  # confirm the fixed target
+sudo systemctl unmask <svc> && sudo systemctl restart <svc>
+curl -s <health>                                      # then a REAL completion
+```
+
+Deploy caveat: "healthy after Ns" = systemd ACTIVE, not model LOADED —
+confirm with a real `/v1/chat/completions`. RTX 3090 (Ampere) has no
+FP8/FP4 tensor cores: FP8 weights run via upcast but must still fit; choose
+quant by size fit, then accept the compute path the card supports. Before
+choosing a smaller quant, state WHICH of the two you are buying — headroom,
+or a design change (a smaller quant does not change what the runtime can
+express; per-tensor cross-device placement stays refused under
+`--override-tensor`, and a layer split moves only contiguous layer ranges).
+
+## 2. Ground truth first
+
+- Read the HF model card (runtime pin, hard constraints, recommended args)
+  + the repo trio `scripts/<svc>.env`, `scripts/serve-<svc>.sh`,
+  `scripts/<svc>.service` + the runbook under `serverAppConfigs-docs/LLM/`.
+- Check the host before downloading anything: `ls /home/dank/models/`,
+  `df -h /home`, `du -sh` — bench-era or prior-deployment files often already
+  exist.
+
+## 3. Fetch and verify (HF payloads)
+
+**Inventory by QUANT NAME, never by extension** — large quants ship as
+`.partNofM` fragments (Q4_K_M = 37.00+37.00+36.97 GiB), so an
+`endswith(".gguf")` filter makes Q4_K_S look nonexistent. Check both
+mradermacher repos (`…-i1-GGUF` imatrix, `…-GGUF` plain) and the model's own
+repo; the tree API's `blobs=true` gives `lfs.oid` per file. Gated mirrors
+return 403 even with a valid token until the account accepts the gate —
+prefer the ungated copy.
+
+```bash
+# token out of band: mode-600 curl config; a token passed as -H is world-readable
+printf 'header = "Authorization: Bearer %s"\n' "$(cat ~/.cache/huggingface/token)" > /tmp/hf.curlrc
+chmod 600 /tmp/hf.curlrc
+curl -fL -C - --retry 10 --retry-delay 5 --retry-max-time 21600 -K /tmp/hf.curlrc -o "$DEST/$name" "$BASE/$name"
+```
+
+`-f` is MANDATORY: `curl -o FILE URL` **exits 0 on HTTP 401/403** and writes
+the error body as the artifact (identical 153-byte shards, cheerful DONE,
+zero bytes of model). Never `pgrep -a` a running fetch (token is in
+`/proc/<pid>/cmdline`); watch progress by size:
+`while [ "$(stat -c%s … || echo 0)" -lt "$EXPECTED" ]; do sleep 30; done`.
+
+**Verify, then assemble:** per part, byte count == Hub's declared size AND
+`sha256sum` == `lfs.oid`; concatenate in NUMERIC order; assembled total ==
+sum of parts (compute the sum with python, not by hand) and `head -c 4` ==
+`GGUF`. The assembled file has no published hash — the per-part oids + exact
+total are the chain. Keep the fragments when disk allows (only copies
+re-verifiable against the Hub); failure artifacts and the replaced payload
+go to `~/.AGENTTRASH/YYYYMMDD-topic`, never `rm` — upstreams DELIST quants
+and the local file may be the last copy.
+
+Fetch-script one-liners: dedicated scripts (`scripts/download-*.sh`) already
+do `-f` + byte-count + LFS sha256 + `--verify-only`; ad-hoc fetches tend to
+skip all three. HF cache snapshots SYMLINK every file into `blobs/` and
+those links dangle inside containers — serve scripts and the deployer refuse
+symlinks and missing tokenizer files for exactly this reason.
+
+## 4. Runtime build (when the card pins a fork/build)
+
+- Clone + build with the card's exact cmake line, plus
+  `-DLLAMA_BUILD_WEBUI=OFF` — WebUI asset provisioning (npm / HF llama-ui
+  bucket) aborts llama.cpp link at 98 % on these hosts; the API server is
+  unaffected.
+- Verify every flag the serve script passes exists in the new binary BEFORE
+  touching the service: `llama-server --help | grep -q -- "<flag>"` for each;
+  list MISSes. Read the GGUF's tensor histogram first if the payload uses a
+  custom type — the FORMAT picks the runtime ([[GGUF Compatibility]]).
+
+## 5. Config via git only
+
+Edit `scripts/<svc>.env` (paths, LLAMA_BIN, MODEL_FILE, MMPROJ_PATH) + serve
+script header/defaults locally; `bash -n`; `./tests/run.sh` must pass;
+commit; push; `git pull --ff-only` on the target. Never edit configs on the
+remote. Renaming a model dir (`mv`) is host state and allowed. The env-file
+hook pattern (`FRANK_*_ENV_FILE`) is what makes bench windows and staged
+arms possible — keep it intact.
+
+## 6. Live validation
+
+- No scratch ports — one GPU cannot host two big models. Ensure the old unit
+  is truly gone first: `systemctl is-active` can hang in `stop-sigterm`
+  (llama.cpp never preempts an in-flight generation); poll until the PID
+  exits.
+- Manual serve for smoke: `setsid nohup bash scripts/serve-<svc>.sh >
+  /tmp/x.log 2>&1 &` — plain `ssh 'nohup … &'` dies with its ssh session.
+- Verify: health 200; a REAL `/v1/chat/completions`; reasoning flows
+  (`reasoning_content` non-empty); if the endpoint takes images, a
+  solid-color probe through an `image_url` content part (legacy top-level
+  `images` is silently ignored; probe with ≥256 px).
+- vLLM specifics: first request after boot can take ~18 s (deferred Triton
+  loads) — warm and discard; `--reasoning-parser` / tool parsers are part of
+  the contract ([[vLLM Boot Forensics and Config Interrogation]]).
+
+## 7. systemd handoff — THE TRAP
+
+`./scripts/deploy.sh <host> --service <svc>` gates health, but a leftover
+manual server holding the port makes the health check pass while the UNIT
+crash-loops on bind. Always confirm ownership:
+`systemctl show <svc> -p MainPID` equals the single `pgrep -x llama-server`
+(or the container runtime PID); only then trust "healthy". If a real
+long-context task is queued, do the handoff anyway — client aborts are
+recoverable, drift is not.
+
+## 8. Close-out (mandatory)
+
+1. Docs: active runbook updated (build recipe, profile, validation numbers
+   incl. real-workload prefill/decode); superseded runbook gets a RETIRED
+   banner and stays as the rollback record; `LLM/index.md` mermaid + active
+   list; AGENTS.md fleet table + runtime section + history chain; work-log
+   entry in `serverAppConfigs-docs/log.md`.
+2. `~/.omp/agent/models.yml`: `contextWindow` must equal the REAL per-slot
+   context — llama.cpp logs `n_ctx_slot` per prompt (`--parallel N` divides
+   the pool; e.g. 262144 native / 2 slots = 131072 per slot); vLLM reports
+   `max-model-len` and its concurrency line. Overselling causes silent
+   client-side overflow or a discovered-wall 400. Also `maxTokens` vs the
+   engine's cap, `input` capabilities (text/image), and
+   `reasoningContentField` per runtime (llama.cpp/SGLang emit
+   `reasoning_content`, vLLM emits `reasoning`). Fix every `modelRoles` entry
+   that named the old endpoint.
+3. **Rollback = one env commit** (paths + LLAMA_BIN + flags), prior binary +
+   payload kept on disk, rollback numbers recorded in the runbook.
+
+## 9. Revert-pin an ad hoc experiment (the three pins)
+
+A campaign is "done" only when all three exist — without them a
+`podman pull`, an image update, or a lost scratch script destroys the way
+back:
+
+1. **Git tag (annotated, pushed):**
+   `git tag -a <exp>-tuned-YYYYMMDD -m "<tuned env/flags + measured headline +
+   runbook path + revert command>"` — a future agent must reconstruct the
+   campaign from the tag message alone.
+2. **Image digest in the serve script:** floating tags move. Capture
+   `podman inspect <ctr> --format '{{.ImageName}} {{.Image}}'` + `podman
+   images --digests`, patch `IMAGE=` to `repo@sha256:<repodigest>` with a
+   comment (image ID, date, old ref), verify `podman image exists
+   repo@sha256:<digest>` and `bash -n`. Digest refs survive upstream pushes;
+   `podman pull repo@sha256:<digest>` re-fetches exact bytes. One host can
+   hold MULTIPLE images of one repo — pin by the digest of the RUNNING
+   container, never by tag name. Never edit the pinned line back to a
+   floating tag; new experiments branch from the digest.
+3. **Local alias:** `podman tag <id> <exp>-runtime:tuned-YYYYMMDD`.
+
+Record scratch-script paths + full flag surface in the runbook (hook env
+vars in CAPS); expired images/scripts go to `.AGENTTRASH`, never `rm`.
+Pure measurement sessions that changed no state need only pin 1.
+
+## 10. Research verification discipline (for "can model X run here?")
+
+- Scouted research summaries have shipped fabricated citations on this fleet
+  (invented HF repos, plausible-but-wrong PR numbers, stale engine versions).
+  NEVER copy a scout's citations into a deliverable — re-verify each
+  load-bearing URL directly (GitHub issues/PRs/repos, HF pages) before it
+  enters anything.
+- The in-repo runbooks are FIRST-HAND evidence and outrank any web claim
+  about our own hosts.
+- Verified anchors (2026-09-27): DeepSeek V4.1-Flash (552B/8B-16B active,
+  CED arch, 1M ctx, MIT; preview was 284B/13B with native FP4+FP8 — early
+  GGUF conversions garbage); Lucebox = the proven gfx1201 ROCm 7.2 path
+  (heterogeneous R9700+Strix-Halo V4.1: 86 tok/s decode, 788 t/s prefill@2K).
+  GLM-5.3-Flash (320B/18B, KDA+DSA+mHC, 1M ctx — ALWAYS bind `-c`, the
+  default hard-hung a 128 GB host; native FP8 328.3 GB; `-fa off` MANDATORY,
+  flash-attn breaks MLA correctness; llama.cpp arch `glm5next` lived in
+  unslothai branch → ggml PR #27754, MTP landed d07e71ede 2026-08-30; quant
+  ladder UD-IQ1_S 93.1 GB → UD-Q4_K_XL 199.7 GB, GSQ-RCO 3.5-bit −1.40 pp
+  MMLU-Pro vs Q8_0).
+- GGUF parsing gotcha: expert tensors are
+  `blk.N.ffn_{gate,up,down}_exps.weight` — 4 dot-parts INCLUDING `.weight`;
+  a `len(parts) == 3` filter matches zero layers with no parse error. Filter
+  on `parts[0]=="blk" and parts[3]=="weight" and "exps" in parts[2]`.
+
+## 11. Adding a second LLM endpoint to a host
+
+When a host must serve two models at once (an iGPU runtime plus a second
+card), or one endpoint is moving and a new one takes the freed hardware.
+
+**0. Decide the shape before touching hardware.** One provider per
+HOST:PORT in `~/.omp/agent/models.yml` (on the box where the harness runs)
+— a second endpoint is a new provider block, not a second model in the
+existing one. Read the budget as an inequality: for unified memory the
+engine's own fit check is the arbiter (`budget = host − pinned weights −
+reserve`; `need = KV + arena + vision`); for a discrete card, weight the KV
+by counting only full-attention layers — and remember `--parallel N`
+DIVIDES `--ctx-size`. A co-tenant is a budget line, not a footnote: free
+RAM by changing the POOL or the ARENA (the reserve is a fit-check term,
+never an allocation) and drop optional charges (a vision tower is ~2 GiB)
+rather than hoping.
+
+**1. Repo first, and retire in the same commit.** New service files
+(`scripts/<svc>.env`, `.service`, `serve-<svc>.sh` with hard guards for
+missing artifacts); planner entries (`SERVICE_ORDER` + the target's service
+set + path mappings, inventories in `deploy.sh`/`deploy-remote.sh`, the
+dispatch case); retire the obsoleted pieces into `.AGENTTRASH/` with a
+suffix — and check what pointed at them: bench harnesses, a chat template
+only the old runtime read, and `scripts/test-changed.sh` mappings are the
+usual victims (a mapping to a retired test file fails the pre-commit
+wrapper silently for weeks). One bats file per service asserting
+INVARIANTS (`pool == 2 × ctx`, KV bytes ≤ card budget, port separation),
+not constants; a negative grep must strip comments first
+(`grep -vE '^\s*#' file | grep -q …`) or it matches the prose explaining
+the very thing you removed.
+
+**2. The port-bind race (rootless podman).** pasta binds the published port
+at container CREATE; a just-stopped predecessor still holds it. Order the
+handler: retire the old manager's unit → stop THIS service's own unit
+(no-op on first cutover) → wait for every socket on the port to clear
+([[Container Host Reachability]]) → install → start. Without the second
+step a redeploy deadlocks against its own gate.
+
+**3. Wire the surfaces, refuse impossible signals.** Point the harness
+small/fast + image roles at the new endpoint; move the vision role WITH the
+vision tower (one owner per capability). `shared/homepage/services.json`: a
+card per endpoint, probed from the monitor's own host. Prometheus: only add
+a job if the runtime actually exposes Prometheus text (halogen publishes
+`/health`, `/cache`, `/v1/*` and nothing else — the job was REMOVED with the
+reason written down rather than left as a permanent red).
+
+**4. Cutover, then verify by driving it.** `deploy.sh <target> --service
+<large> --service <small>` (order matters); per endpoint: `/health`,
+`/v1/models` reports the alias, a chat completion returns content AND
+`reasoning_content`, a `tools` request returns `tool_calls`, an image
+request is answered (or refused with a named reason), the runtime's own
+speculation counters move, and the unit is `enabled`/`generated` with linger
+on. Read the engine's boot ledger for a silent clamp. Finally spend one
+real request through the harness per changed role — routing labels lie; the
+servers' own journals do not.
+
+Traps that look like model faults: `500 tools param requires --jinja flag`
+(runtime lacks the Jinja path — pin it); `500 Jinja Exception: Unexpected
+reasoning effort <level>` (fold via `compat.reasoningEffortMap`); the same
+request works by hand but a subagent dies (subagents run at a different
+thinking level — check both); the container dies instantly although the
+predecessor "stopped" (the port race above).
+
+## Related
+
+- [[GGUF Compatibility]] — which runtime can load a payload at all
+- [[Llama Cpp Serving Sizing]] — pricing the shape once it loads
+- [[RDNA4 MoE Expert Offload VRAM Budget]] — vLLM-side fit arithmetic
+- [[LLM Bench Windows and Measurement]] — measuring the new deployment
+- [[dankserver-qwen38-fp8-expert-offload-tuning]] — the current endpoint

serverAppConfigs-docs/LLM/PCIe Link and Atomics.md

@@ -0,0 +1,244 @@
+---
+type: reference
+title: PCIe Link and Atomics
+description: Link probing (idle downshift, path ceilings, CUDA/HIP bandwidth probes), the streaming-vs-compute ceiling arithmetic, and the PCIe-atomics wall that decides whether AMD multi-GPU collectives can exist at all.
+tags: [pcie, rccl, atomics, multi-gpu, bandwidth, amd, nvidia]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# PCIe Link and Atomics
+
+## 1. Degraded or just idle?
+
+The trap: an idle GPU reports `LnkSta: Speed 2.5GT/s (downgraded)` — the
+driver downshifts to Gen1 for power. That is not a fault and not the slot's
+fault. Never conclude from an idle reading.
+
+1. **Path ceiling from the kernel (cheapest, most decisive):**
+   `sudo journalctl -k -b --no-pager | grep "available PCIe bandwidth"` →
+   `pci 0000:c4:00.0: 63.012 Gb/s available PCIe bandwidth, limited by 16.0
+   GT/s PCIe x4 link at 0000:00:03.1 (capable of 252.048 Gb/s with ... x16)`.
+   The "limited by" device names the **bottleneck** (usually a root port) and
+   its width — that is the number a swap could move; "capable of x16" is the
+   endpoint's OWN capability, and the gap between them is the story.
+   `dmesg` rotates; `journalctl -k -b` keeps this boot.
+2. **Both sides separately:** `LnkCap`/`LnkSta`/`LnkCtl2` on endpoint AND root
+   port. `LnkCtl2: Target Link Speed: 16GT/s` on both ends = the platform
+   intends Gen4, so a Gen1 current speed is power management, not
+   misconfiguration. A root port whose `LnkCap` width exceeds the trained
+   width means the socket/adapter wires fewer lanes — **M.2 connectors are 4
+   lanes by pinout; an M.2 riser can never deliver more.**
+3. **Prove the link under load (the step people skip):** on NVIDIA sample NVML
+   while the GPU works
+   (`nvidia-smi --query-gpu=pcie.link.gen.current,pcie.link.gen.hostmax,pcie.link.width.current,pcie.link.width.max`):
+   `gen.current` climbing 1 → `hostmax` = idle downshift, healthy; stuck below
+   `hostmax` under load = real degradation; `width.current < width.max` under
+   load = **physical lane limit** — no software touches it. On a busy
+   production GPU this is a usage smoke test, not a benchmark — never stop a
+   live service to read its link.
+4. **Do not plan hardware from `dmidecode -t slot`** — consumer firmware
+   labels Ethernet/SD controllers as "M.2 slots" and advertises slots in PCI
+   domains that do not exist. Trust `lspci -nn | grep "PCI bridge"` and
+   `lspci -tv` (empty root ports still enumerate; an absent port is absent).
+5. **"Would moving the card help?"** Compare the bottleneck device from step 1
+   with the target port's `LnkCap` width; two M.2 sockets = both x4 = a
+   no-op. Clean `DevSta`/`CESta` with a low trained speed means "wired
+   narrow", not "failing". Report ceiling, card capability, what the idle
+   reading meant, and whether any physical change moves the number.
+
+Gen4 usable payload after 128b/130b: **x4 7.877 GB/s, x8 15.75, x16 31.5**.
+
+## 2. The engine shape decides whether the link matters at all
+
+| engine shape | what crosses PCIe per token | link matters? |
+|---|---|---|
+| streams weights (expert LRU, paging) | the missed weights (~GB/token) | **yes — decisive** |
+| computes offloaded weights in RAM (llama.cpp `-ot exps=CPU`) | activations only (~KB) | no; host RAM bandwidth decides |
+| everything resident | nothing | no |
+
+**Ceiling arithmetic:** `link time/token = active expert bytes / measured
+H2D rate`; `ceiling = 1 / link_time × resident-fraction saved (LRU hits)`.
+Compare against the compute-in-RAM path at host RAM bandwidth (256 GB/s
+LPDDR5X on the Strix Halo). **If the link is >10x slower than the memory
+behind it (x4 is 31x here), streaming loses by construction and no tuning
+rescues it; doubling the link only doubles a losing number.**
+
+Worked example (Dankclaw 2026-09-16): Gen4 x4 measured H2D **6.72 / D2H 6.60
+GB/s** (85 % of spec) — pinned H2D ≈ async, pageable within 1 %, i.e.
+link-limited not staging-limited. A 3 MiB slab (the expert-streaming
+granularity) = 474 µs. A3B MoE at Q4_K_M (~1.70 GiB active/token) → 272
+ms/token ⇒ **3.7 tok/s** (4.6 with 20 % resident) vs measured **20.46 tok/s**
+with experts computed on the CPU, 25–26.5 on the iGPU. The streaming design
+loses 3–5x.
+
+## 3. Measure the link: two probes, same mechanics
+
+Stop **all** production services first (fleet bench law) — the 3090's VRAM
+may be nearly full, and the probe contends for the very link it measures.
+
+**CUDA probe** (`pcie-link-test.cu`): sweep {3, 16, 64, 256 MiB, 1, 2 GiB}
+with `cudaMalloc` + `cudaHostAlloc`, event-timed; per size measure sync H2D,
+async H2D, D2H, one single transfer, and pageable H2D. `3 MiB` is the row
+that matters for expert streaming; big sizes give the ceiling.
+
+**HIP probe for RDNA4** (`pcie_bw.hip`) — the ROCm landmines each cost real
+time, so encode them:
+
+- `hipGetDeviceProperties` SEGFAULTS on some builds; use `hipGetDeviceCount`
+  + `hipSetDevice` only. First line: `setvbuf(stdout, NULL, _IONBF, 0);` — a
+  segfaulting process loses block-buffered stdout and hides the crash point.
+- Build with an explicit arch: `HIP_TARGETS=gfx1201 hipcc -O2
+  --offload-arch=gfx1201 -o p p.hip`; compile with `| grep -E "error:"`,
+  never `| tail -N` (truncation hid a missing `#include <cstring>`).
+- `hipMemset`/`hipDeviceSynchronize` are `[nodiscard]` — use or cast.
+- Mount the whole `/dev/dri` into ROCm containers; individual `renderD*`
+  nodes enumerate but fail context creation.
+- `rocm-smi --showmetrics`: `pcie_link_width` is trustworthy, but
+  `pcie_link_speed` reads the ASPM idle value and `pcie_bandwidth_acc/inst`
+  are usually `N/A` on consumer cards — live traffic generally cannot be
+  sampled.
+
+```cuda
+// pcie_bw.hip — sweep sizes, pinned vs pageable
+#include <cstdio>
+#include <cstring>
+#include <cstdlib>
+#include <hip/hip_runtime.h>
+static double one(size_t bytes, bool pinned, int iters) {
+    void *dev = NULL, *host = NULL;
+    if (hipMalloc(&dev, bytes) != hipSuccess) return -1;
+    if (pinned) { if (hipHostMalloc(&host, bytes, hipHostMallocDefault) != hipSuccess) return -1; }
+    else        { host = malloc(bytes); }
+    memset(host, 0x5a, bytes);
+    hipEvent_t a, b; hipEventCreate(&a); hipEventCreate(&b);
+    for (int i = 0; i < 3; i++) hipMemcpy(dev, host, bytes, hipMemcpyHostToDevice);
+    hipDeviceSynchronize();
+    hipEventRecord(a, 0);
+    for (int i = 0; i < iters; i++) hipMemcpy(dev, host, bytes, hipMemcpyHostToDevice);
+    hipEventRecord(b, 0); hipEventSynchronize(b);
+    float ms = 0; hipEventElapsedTime(&ms, a, b);
+    double gb = (double)bytes * iters / (ms / 1000.0) / 1e9;
+    if (!pinned) free(host); else hipHostFree(host);
+    hipFree(dev); hipEventDestroy(a); hipEventDestroy(b);
+    return gb;
+}
+int main(int argc, char **argv) {
+    setvbuf(stdout, NULL, _IONBF, 0);
+    if (hipSetDevice((argc > 1) ? atoi(argv[1]) : 0) != hipSuccess) return 1;
+    size_t sz[] = {2459136UL, 3UL<<20, 16UL<<20, 64UL<<20, 256UL<<20, 1UL<<30};
+    for (int i = 0; i < 6; i++) {
+        size_t b = sz[i];
+        int iters = (b >= (1UL<<30)) ? 10 : (b >= (256UL<<20)) ? 20 : 100;
+        printf("  %10.3f MiB | pinned %6.2f | pageable %6.2f\n",
+               b/1048576.0, one(b,true,iters), one(b,false,iters));
+    }
+    return 0;
+}
+```
+
+**Reading it:**
+
+- **Engine-vs-link is the diagnosis, not the raw number:** compare
+  `engine_effective_GB_s` to `link_GB_s` at the SAME transfer size. Engine at
+  ~40 % of the link at the same granularity = the wire is not the problem and
+  no hardware change will help.
+- Achieved GB/s is itself the proof of the trained link state (6.7 GB/s only
+  exists at Gen4 x4) — no separate under-load `LnkSta` sample needed.
+- **pinned ≈ pageable ⇒ pinning is not a lever.** async ≈ sync ⇒ link-limited
+  (the usual "pageable costs 2x" rule does not apply at this link).
+- Small-transfer penalty (2.345 MiB: ~106 µs vs ~78 µs ideal, ~21 %) is
+  per-transfer overhead — recoverable only by BATCHING transfers, not by a
+  faster link. (Dankserver 2026-09-29: bulk 22.7 GB/s at the 2.344 MiB slab =
+  91 % of Gen5 x8, while the engine's effective read path ran ~9.5 GB/s — the
+  gap is the fork's gather software path, not the wire.)
+- Both GPUs symmetric ⇒ no cross-device rebalance is available; measure both.
+- Don't quote a derived line from inside a probe without hand-checking it —
+  a printf format/argument mismatch in one of these printed a nonsense
+  ceiling. GiB→GB before dividing.
+
+## 4. The PCIe-atomics wall (AMD multi-GPU collectives)
+
+**Symptom:** multi-GPU AMD host fails TP/PP; the failure surfaces as
+`hipLaunchKernel: hipErrorIllegalState` for `ncclDevKernel_*` — nothing in
+vLLM/RCCL/PyTorch names atomics. Get the real message:
+
+```bash
+AMD_LOG_LEVEL=3 <the failing process>
+# → "Pcie atomics not enabled, hostcall not supported" → "AQL dispatch failed!"
+```
+
+**Cause chain:** RCCL kernels carry a hidden hostcall-buffer parameter; ROCr
+refuses to fill it without PCIe atomics, which it reads **once at driver
+probe** (`rocclr/device/rocm/rocdevice.cpp` `checkAtomicSupport()` — no else,
+no override). **No env var, module param, or sysfs toggle exists** —
+`setpci -s <gpu> CAP_EXP+28.w=0440` (DCTL2 bit 6) changes nothing at runtime.
+
+**Check the CHAIN, not the card:**
+
+```bash
+sudo lspci -vv -s <gpu> | grep -E "AtomicOpsCtl|LnkSta|LnkCap"  # want ReqEn+
+lspci -tv                                                        # every hop
+sudo lspci -vv -s <bridge> | grep AtomicOpsCap                   # want Routing+
+```
+
+`Routing-` is a read-only hardware capability. Typical AM5 layout: one x16
+slot on CPU root ports (atomics fine), the rest behind the chipset (usually
+not). Bandwidth for TP is small anyway: ~10 MB per token-step for a 27B at
+batch 4 ≈ 1.4 ms at Gen4 x4 — compute-bound, not interconnect-bound.
+
+**Probe matrix (bare 2-rank `torch.distributed`, one process per GPU, 25 s PG
+timeout) — one probe answers several design questions:**
+
+| Op | CPU-lane card | chipset card | Decides |
+|---|---|---|---|
+| `broadcast`/`all_reduce`/`barrier` | OK | FAIL | TP is dead |
+| `send`/`recv` (P2P), in ISOLATION | hang | FAIL | **PP is dead too** — vLLM's PP handoff IS this primitive |
+
+Test `send`/`recv` alone (a prior failed op leaves the group broken). Run the
+probe from a REAL file (`torch.multiprocessing.spawn` re-executes it; piping
+into `python3 -` fails with `FileNotFoundError: '/<stdin>'`). `world_size=1`
+always succeeds — **one-card success proves nothing.**
+
+**SAFETY — non-negotiable:** never run a GPU experiment beside a serving
+unit: `sudo systemctl stop <svc> && sudo systemctl mask <svc>` first (mask,
+never disable — disable deletes a relinked unit file), or use a host that
+serves nobody. A hung AQL dispatch can wedge the GPU driver; a probe beside a
+live service on 2026-09-16 preceded a hard host down (no ARP, no ICMP from
+two subnets, power cycle required). Telling off from wedged afterwards: `ip
+neigh show <ip>` from a peer — `FAILED`/`INCOMPLETE` = off/NIC dead; a wedged
+kernel usually still answers ARP. Validate your own probe's naming before
+reporting an outage (`gitea` vs `gitea-server`).
+
+**Fixes, ranked:**
+
+1. CPU-attached M.2 → PCIe x4 riser (~$30) — puts the card on the CPU root
+   complex; x4 suffices for TP.
+2. Board whose CPU x16 bifurcates x8/x8 (done 2026-09-20 on dankserver — both
+   R9700s on root ports `00:01.1`/`00:01.3`, every bridge `Routing+`, both
+   endpoints `ReqEn+`).
+3. llama.cpp `--split-mode layer` — no collectives at all ([[Llama Cpp Multi-Device Splits]]); costs the GGUF conversion plus paged attention /
+   continuous batching / spec decode.
+4. Independent TP=1 replicas + routing proxy — cards never share one model.
+
+**Closed software routes (do not re-litigate):** vLLM's IPC custom allreduce
+is gated by a CUDA-sm-keyed table (`{9.0, 10.0, 10.3, 10.7}`; ROCm reports
+12.0 → off on RDNA4 by code limit) — and patching it is insufficient because
+TP init's `broadcast`/`barrier`/logits `all_gather` have no non-RCCL path. A
+fork wrapping `.all_reduce` (e.g. radiance `RADIANCE_USE_R4D_AR` one-shot
+P2P, gloo for IPC-handle exchange) does not intercept
+`PyNcclCommunicator.__init__`'s warmup allreduce. NCCL symmetric memory needs
+`ncclDevComm*` (many builds ship the version string without the device API).
+P/D disaggregation needs a KV connector AND coordination collectives. gloo is
+CPU/TCP — unreachable for per-layer TP.
+
+**Open, unverified:** rebuilding RCCL so kernels stop declaring a hostcall
+buffer (`-mprintf-kind=buffered` is the usual claim) — counter-evidence: no
+live `printf`/`assert` in RCCL device source. Test path: rebuild, dump a
+kernel's declared args (`roc-obj-extract` + `llvm-readelf --notes`), look for
+`hidden_hostcall_buffer`; if gone, retry the 2-rank probe (service stopped).
+
+## Related
+
+- [[Llama Cpp Multi-Device Splits]] — the layer-split path around the wall
+- [[Dual-R9700-Cooperation]] / [[x4-pair-experiment]] — measured history
+- [[dankserver-qwen38-fp8-expert-offload-tuning]] — the live gather economics

serverAppConfigs-docs/LLM/RDNA4 Container Serving.md

@@ -0,0 +1,250 @@
+---
+type: reference
+title: RDNA4 Container Serving
+description: Bringing a model larger than VRAM onto Dankserver's 2x R9700 (gfx1201) with the tcclaviger vLLM stack — quant-class gating, fetch verification, container GPU-init triage, memory budgeting and the expert-offload plan.
+tags: [rdna4, gfx1201, r9700, vllm, expert-offload, podman, rocm]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# RDNA4 Container Serving
+
+Bring-up path for Dankserver (2x Radeon AI PRO R9700, gfx1201, 31.86 GiB each =
+**63.7 GiB real VRAM**, 246 GiB RAM, 7700X) with
+`docker.io/tcclaviger/vllm` (vLLM 0.29.0.dev, GLM/Qwen-specialised RDNA4 build
+with native MoE quant kernels and a first-class expert offload planner). The
+tuning history and current deployed shape live in
+[[dankserver-qwen38-fp8-expert-offload-tuning]]; this page is the bring-up law.
+
+## 0. Know the box — and its phantom third GPU
+
+`torch.cuda.device_count()` returns **3**; summing `total_memory` overstates
+the box as 186.7 GiB. The third device is the 7700X's integrated Vega
+(gfx1036): torch reports ~123 GiB for it, `rocm-smi` reports 0.50 GiB,
+allocating on it moves system `MemAvailable` 1:1 (it is GTT over DDR5, not
+dedicated VRAM), and it dies with `hipErrorInvalidKernelFile` when the image
+ships no code for that arch. **Never sum device `total_memory`; never count
+the iGPU as headroom or a workload.**
+
+## 1. Preflight: is the quant class EXECUTABLE, not merely present?
+
+Expert offload is mandatory when weights exceed VRAM, and it requires the
+checkpoint's experts class to have BOTH a compiled kernel AND an
+`offload_recipe` ("a pinned expert has no host copy, so there is no
+read-through fallback" — startup aborts otherwise). Check BEFORE downloading
+anything:
+
+```bash
+podman run --rm --device /dev/kfd --device /dev/dri --device /dev/mem \
+  -e HIP_VISIBLE_DEVICES=0,1 --entrypoint bash $IMAGE -c '
+  nm -D /app/r4dhip/r4d.so | grep -oE "gemm_moe_[a-z0-9_]+" | sort -u
+  grep -rl offload_recipe /opt/vllm/lib/python*/site-packages/vllm/model_executor/layers/fused_moe/'
+```
+
+| quant | kernel in `r4d.so` | experts class | verdict |
+|---|---|---|---|
+| FP8 | yes | `Fp8HipMoEExperts` | **serves** |
+| MXFP4 / DSFP4 | yes (`gemm_moe_mxfp4a8_*` / `dsfp4a8_*`) | `R4dMxfp4MoEExperts` / `R4dDsfp4MoEExperts` | **serves** |
+| W4A16 | **no** `gemm_moe_w4a16_*` | `R4dW4a16MoEExperts` exists | silently falls back to generic `TritonWNA16Experts`, no recipe → startup halts |
+| RFA | none | no class at all | out |
+
+Also confirm the architecture routes natively with NO overlay
+(`ModelRegistry.get_supported_archs()` for `Qwen4Exp`/`Glm5`). The stock image
+natively carries Qwen3.8/Qwen4Exp and GLM-5.x including MTP heads. **Do not
+apply vendor overlays** against a fast-moving pre-release image — the breakage
+misleads (a missing symbol in the overlay's OWN copy of a module masks the
+real cause, then moves to another symbol each patch;
+`cannot import name 'checkpoint_has_lm_head'` = overlay clobbered
+`transformers_utils/config.py`). The image ships quantizers for
+`rfa|rfi|mxfp4|mxfp4ocp|paro5` under `/app/tools/quantize.py`
+(`--fp8-attn --quant-only --dry-run` first) if a re-quant is needed — note it
+can run MXFP4/DSFP4/FP8 but only PRODUCE rfa/rfi/mxfp4/paro.
+
+## 2. Fetch with verification, always
+
+Gated repos: token never in argv (`pgrep -a curl` prints command lines).
+Build a mode-600 curl config once:
+
+```bash
+{ printf 'header = "Authorization: Bearer '; cat ~/.cache/huggingface/token; printf '"\n'; } \
+  > ~/.cache/huggingface/curl.conf && chmod 600 ~/.cache/huggingface/curl.conf
+curl -K ~/.cache/huggingface/curl.conf -f -sS -L -C - -o "$out" "$url"
+```
+
+`--fail` is MANDATORY — `curl -o FILE URL` exits 0 on HTTP 401/403 and writes
+the error body as a "successful artifact" (a gated repo once yielded three
+153-byte shards with identical sha256s and a cheerful DONE banner). `-C -`
+resumes torn shards; HTTP/2 cancels leave silently truncated files, so
+re-verify EVERY shard's byte count against the Hub manifest (and the LFS
+sha256) before serving. The AGENTS.md Secret Rules carry the token-hygiene
+law; [[frank-medium-model-swap]] has the same detail for GGUFs and
+`.partNofM` fragments.
+
+## 3. Serve — the known-good invocation and its traps
+
+```bash
+sudo sh -c 'sync; echo 3 > /proc/sys/vm/drop_caches'
+podman run -d --name <ctr> --ipc=host --ulimit memlock=-1:-1 \
+  --cap-add SYS_PTRACE \
+  --device /dev/kfd --device /dev/dri --device /dev/mem \
+  -e HIP_VISIBLE_DEVICES=0,1 -e VLLM_ROCM_USE_AITER=0 \
+  -v /home/dank/models/<model>:/models:ro \
+  $IMAGE /models --served-model-name <name> --port 7770 \
+  --tensor-parallel-size 2 --max-model-len 65536 \
+  --kv-cache-memory-bytes 2147483648 \
+  --max-num-seqs 4 --gpu-memory-utilization 0.95 \
+  --enable-prefix-caching --trust-remote-code
+```
+
+- **Entrypoint already execs `vllm serve "$@"`** — pass the model path as the
+  first argument and do NOT pass `serve` (podman then tries to pull `serve`
+  as an image name). `-e VAR=1` must come BEFORE the image name.
+- **Mount the whole `/dev/dri`, never individual render nodes** (§4).
+- `--ipc=host` and `--shm-size` are mutually exclusive in podman.
+- `podman run -d` without `-a` sends container stdout nowhere — read
+  `podman logs <ctr>`; and vLLM writes `Application startup complete` to
+  **stderr**, so a `podman logs X 2>/dev/null | grep -q` readiness gate never
+  fires.
+- Start at 65536, not the native ceiling; prove the plan, then scale.
+- Load takes ~5 min with `/health` refusing the whole time — do not kill the
+  process (deploy.sh waits 20 min).
+
+## 4. Container GPU-init triage (ordered)
+
+1. **Enumerate-vs-context trap (most common):** individual
+   `renderD*` nodes give HIP the render nodes WITHOUT the primary `card*`
+   nodes — it enumerates agents (`device_count()` returns N) and then refuses
+   every context: `RuntimeError: No CUDA GPUs are available` on the first
+   real op. A/B-confirmed 2026-09-19. Rule: `--device /dev/kfd --device
+   /dev/dri --device /dev/mem`; restrict GPUs with `HIP_VISIBLE_DEVICES`,
+   never with device nodes.
+2. **`device_count()` is not evidence.** Prove usability with a real op
+   (`(a@a).sum().item()` on a fresh `torch.randn`), then repeat via
+   `mp.spawn`, `mp.fork`, and `subprocess.run([sys.executable, ...])`, and
+   cross-check `rocm-smi --showmeminfo vram`.
+3. **memlock rlimit → a MISLEADING CUDA OOM.** Host default 8192 KB; a pin
+   larger than the limit fails as
+   `torch.AcceleratorError: CUDA error: out of memory` raised from
+   `pin_memory()` — it names the accelerator but the fault is host pinned
+   memory. It cannot be raised from inside a container or even from root's
+   own shell (`Operation not permitted`); only PAM applies it at login, so
+   fixes only affect NEW sessions:
+   `/etc/security/limits.d/99-memlock.conf` with
+   `* - memlock unlimited` + `root - memlock unlimited`; the container needs
+   `--ulimit memlock=-1:-1` to inherit. Discriminating test after the fix:
+   the container reports `unlimited` AND small pins succeed; if pins still
+   fail, memlock was never the cause.
+4. **`/proc/meminfo` TAB trap:** the key/value separator is a TAB, so
+   `grep MemAvailable /proc/meminfo | tr -s " " | cut -d" " -f2` returns
+   EMPTY and a following awk prints `0.0` — a false "host out of memory" on a
+   host with hundreds of GiB free, which invites aborting something healthy.
+   Tab-safe: `awk -v k='MemAvailable:' '$1==k {printf "%.1f", $2/1048576;
+   exit}' /proc/meminfo`, cross-check `<= MemTotal`, refuse to print a zero.
+   Use `scripts/dankstatus.sh` instead of hand-rolling.
+5. **Silent-parse traps:** `grep -c PATTERN file || echo 0` yields TWO lines
+   (grep prints 0 AND exits 1) — `[: 0\n0: integer expected`; use
+   `| head -1` or `|| true`. `systemctl is-enabled` can emit multiple lines.
+   `rocm-smi --showmeminfo vram --csv` column order is `device, VRAM Total
+   Memory (B), VRAM Total Used Memory (B)` — reading $3 as total/$4 as used
+   yields 0.00; filter to `^card[0-9]+$`.
+6. **Guard the box:** long loads with co-tenants can drive RAM to 0 — abort
+   before the OOM killer (`MemAvailable < 4 GiB → podman rm -f`, exit 1).
+7. **Drop page cache before a large load** (`echo 3 > /proc/sys/vm/drop_caches`)
+   and confirm the drop was actually in effect for the run before treating a
+   failure as a capacity wall — a stale "it doesn't fit" from a contaminated
+   run is expensive to unwind.
+
+## 5. Memory budgeting: the levers are not where they look
+
+The planner builds `budget = total × util`, subtracts a KV reserve, and gives
+the REMAINDER to the expert arena. Consequences, all measured:
+
+- **`--gpu-memory-utilization` grows the ARENA, not the KV pool** (KV stayed
+  0.64 GiB at util 0.95 and 0.98). The arena is the only elastic term.
+- **Lowering `--max-model-len` is the WRONG lever and backfires:** a smaller
+  context shrinks the reserve, handing the space to the arena — measured 65536
+  → 0.64 GiB KV, 32768 → 0.34 GiB KV.
+- **Hybrid-attention undercount:** the planner's per-token KV figure (9190
+  B/token for Qwen4Exp) misses the large fixed per-sequence Mamba state
+  (~2x), so the reserve is always short and the arena silently covers it.
+  Fix by pinning the pool: `--kv-cache-memory-bytes <BYTES>` (2 GiB →
+  113,595 tokens / 3.47x concurrency at 32k). For long context,
+  `--kv-offloading-size` (GiB) spills KV to host.
+- **`--expert-offload-mem` caps the HOST side, not the VRAM arena**
+  (`cap = int(x * GIB)` against `host_memory_available()`). Do not reach for
+  it to fix a VRAM problem; to grow the arena, shrink the KV pin or raise
+  util (see [[dankserver-qwen38-fp8-expert-offload-tuning]] for the
+  util=1.0 unbootability law).
+
+**Read the plan — it is the real acceptance test:**
+
+```
+expert offload: AUTO-ENABLED, weights 62.54 GiB per rank exceed the 30.27 GiB budget (TP2)
+  budget (total x util)      30.27 GiB
+  non-expert weights          6.28 GiB
+  KV reserve                  0.36 GiB   <- sanity-check against the model's own per-token figure
+  CUDA graphs                 0.66 GiB
+  runtime overhead            7.60 GiB
+  expert cache space         15.36 GiB   6712 slots of 2.344 MiB
+  host-backed experts/rank   52.74 GiB   23040 experts
+  host: 105.48 GiB across 2 ranks of 160.89 GiB usable
+```
+
+Host fits when `host_total < usable`. The PLE n-gram table rides memfd and is
+SHARED across TP ranks.
+
+**Loading mechanism** (for reasoning about peak RAM): weights never materialise
+in VRAM — expert Parameters start as CPU staging tensors; each layer's
+`weight_loader` counts shard events and, on layer completion,
+`ExpertTopology.split_layer` moves that layer's experts into the VRAM arena
+and compacts the rest into exact-size pinned host tensors behind UVA. Peak
+host RAM = `host-backed set + layers still loading`, not the whole model. At
+runtime `apply_hot_cold` sends arena-resident experts down a hot GEMM and the
+rest straight over PCIe via UVA views — no staging copy in the critical path.
+
+**Emulation check:** a dequantize-to-compute path doubles real host RAM (118
+GiB offload request → ~238 GiB) and is invisible in the weight count. Read the
+log for `Using 'EMULATION' Mxfp4 MoE backend` before concluding "doesn't fit".
+
+## 6. Performance profile and honest measurement
+
+Decode on this pair is **PCIe-gather-bound, context-independent**: 12.11
+tok/s FLAT from 304 to 16,095 context tokens — KV attention is not the cost.
+With ~29 % of experts arena-resident, ~0.84 GB crosses PCIe per token at the
+engine's effective read rate; cold-vs-warm differing ~10 % confirms the LRU
+works; growth is linear in hit rate, and only speculative decode cuts
+sequential STEPS rather than making each cheaper.
+
+**Benchmark traps that produced fake numbers here:**
+
+- `max_tokens=1` is not TTFT (the server can satisfy it without doing the
+  work — prefill looked flat at 1.26 s for 16k AND 30k prompts). A 64-token
+  non-streamed reply is time-to-64-tokens (decode-bound, flat ~9.6 s at every
+  length). Correct TTFT: stream and time the FIRST content-bearing chunk.
+- A small repeating word loop is prefix-cacheable AND n-gram-predictable — a
+  43-word cycle reported 22,897 tok/s prefill, almost entirely cache hit. Use
+  a large vocabulary in shuffled order and print `cached_tokens`.
+- ~1.2 s fixed per-request overhead swamps prefill below ~16k tokens; compare
+  prefill at 30k+ only.
+- Size prompts with `/v1/messages/count_tokens` (a chat call for counting
+  pollutes the metric); guessing words-per-token silently overran the window
+  (HTTP 400).
+- Always run a canary: plant a unique token at the END of a long prompt and
+  ask the model to quote it — proves prompts reach the model at all.
+- Measure warm AND long (`podman exec` + urllib at 127.0.0.1:7770): a 33-token
+  reply read 14.6 cold / 20.3 warm; a 768-token generation hit 25.7; two
+  concurrent streams scaled to 47 aggregate.
+
+**Levers (all opt-in):** `--expert-prefetch N` takes an INT depth, not a
+boolean (default 0; at depth 1 only ~10 % cold-vs-warm difference — the cache
+already works); `CLAV_TUNABLEOP_SWEEP=1` autotunes GEMM shapes (silently
+skipped under expert offload unless `CLAV_TUNABLEOP_WITH_OFFLOAD=1` too —
+see [[dankserver-qwen38-fp8-expert-offload-tuning]] for why it was rejected);
+`--expert-precise-memory` swaps the fitted overhead figure in.
+
+## Related
+
+- [[dankserver-qwen38-fp8-expert-offload-tuning]] — deployed shape, plan law,
+  measured campaigns
+- [[PCIe Link and Atomics]] — why decode is gather-bound and what cannot fix it
+- [[RDNA4 MoE Expert Offload VRAM Budget]] — arena/plan arithmetic deep-dive
+- [[frank-large-benchmark]] and [[frank-small-benchmark]] — window discipline on live hosts

serverAppConfigs-docs/LLM/RDNA4 MoE Expert Offload VRAM Budget.md

@@ -0,0 +1,219 @@
+---
+type: reference
+title: RDNA4 MoE Expert Offload VRAM Budget
+description: The expert-offload plan arithmetic on the tcclaviger fork — which terms a deploy can move, the util law, the CUDA-graph charge, what free VRAM really is, knife-edge survivability, parallelism ceilings and the context ladder.
+tags: [vllm, expert-offload, vram, arena, kv-cache, ladder, tcclaviger]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# RDNA4 MoE Expert Offload VRAM Budget
+
+For VRAM-bound vLLM on the tcclaviger RDNA4 fork (frank-primary: Qwen3.8-
+Flash-Next FP8, TP=2, 2x R9700 at 31.86 GiB each, expert offload to host RAM).
+Tuning history and the deployed shape live in
+[[dankserver-qwen38-fp8-expert-offload-tuning]]; boot-failure capture lives in
+[[vLLM Boot Forensics and Config Interrogation]].
+
+## 1. The plan law
+
+```
+arena = budget − non_expert − kv_reserve − cuda_graphs − runtime_overhead
+budget = device_total × gpu_memory_utilization          # expert_plan.py build_plan
+slots  = arena // expert_bytes                          # 2.344 MiB/slot ⇒ 437 slots/GiB
+```
+
+**The arena is DEVICE-resident, so the plan predicts reality:** the fp8
+3-window shape's plan totalled 28.27 GiB and `rocm-smi` measured exactly
+28.27 GiB. `weights + pinned_pool + graphs + arena + non-torch (≈1.45 + 0.57
+driver reserve)` vs 31.86 GiB predicts a boot failure before it happens.
+
+Which terms a deploy can move:
+
+| term | reachable? | how |
+|---|---|---|
+| budget (util) | **yes, law-capped** | `--gpu-memory-utilization`; +0.01 = +0.32 GiB on 31.86 GiB cards |
+| CUDA graphs | **yes** | container env `VLLM_CUDAGRAPH_MEM_BASE_MB` / `_PER_GRAPH_MB` (or `_ESTIMATE_MB` for an explicit total) |
+| non-expert weights | no | checkpoint census |
+| KV reserve | no | registry constant: 9190 B/token × max_model_len × 1.08 + per-seq mamba pages |
+| runtime overhead | partly | `--expert-precise-memory` swaps default 7.60 GiB for the registry-fitted 5.25 GiB |
+
+Anything else is an IMAGE EDIT, not a deploy — say so; do not invent a flag
+(verify with `podman exec <ctr> grep -n getenv …/offloader/expert_activation_registry.py`).
+
+**UTIL LAW: 0.982, NEVER 1.00 (measured 2026-10-03).** The CLI field is
+`Field(gt=0, le=1)`, but the worker init demands `free > budget`: at 1.0
+budget == device total and BOTH ranks raise
+`ValueError: Free memory on device cuda:N (31.29/31.86 GiB …)` BEFORE a
+single weight loads — a crash loop (NRestarts ~2 min apart, zero payload
+progress). Not a knife-edge: **unbootable**. Even bootable near-ceiling
+shapes die under real load (0.13 GiB free bf16 boot; `0 bytes free` on a real
+995-token request).
+
+## 2. The CUDA-graph term: charge the measured capture
+
+Defaults (256 MiB base + 52 MiB/graph) over-charge badly. Measure with the
+engine's own post-capture line `Graph capturing finished … took X GiB` (per
+rank; heavier rank binds). Measured on the no-spec piecewise plan: 0.31/0.17
+GiB, and one boot reported a NEGATIVE −0.06 GiB delta (capture fit inside
+pages the profiler had already counted) — the analytic base pool is pure
+over-charge. Shipped: `BASE_MB=0`, `PER_GRAPH_MB=12` = 0.07 GiB — per-graph,
+so a spec-decode shape (18 graphs) still pays (0.21 GiB). An explicit
+`_ESTIMATE_MB` total does NOT scale with ladder changes; base+per-graph does.
+
+**Verify every deploy:** `Estimated CUDA graph memory` line vs the `Graph
+capturing finished` line; if capture exceeds the charge, raise `_PER_GRAPH_MB`
+back toward 52. Check `VLLM_CUDAGRAPH_*` actually present in
+`podman exec … env`.
+
+## 3. The pool is invisible to the plan; free VRAM is the estimator gap
+
+`--kv-cache-memory-bytes` does NOT appear in the arena arithmetic — the plan
+reserves the registry price, not the actual pool. Raising the pool consumes
+real free VRAM without growing the arena; the registry over-prices the pool
+(9190 B/token planned vs 7,891.8 measured at 200704 fp8 — 7,701 at 262144;
+**never reuse a B/token ratio across geometries**, and subtract ~3.87 MB
+block-alignment loss when converting a token target to bytes). The ask lands
+block-aligned: 3,166,500,000 B → 401,408 tokens; verify with
+`GPU KV cache size: N tokens, Maximum concurrency … X.Xx`.
+
+**Driver-reported free VRAM after hours of traffic is NOT allocatable
+arena.** It is the estimator gap: the registry over-prices KV and charges
+5.25 GiB overhead against ~2.5 measured; torch's allocator does not return
+slabs, so it reads as the high-water mark. The only reachable levers are util
+(within the law) and the graph charge; everything else needs a registry edit
+inside the image — the fork's own log line says as much.
+
+**MNBT is arena-free up to the anchor:** registry overhead =
+`base + 0.22 GiB × max(0, (nbt−4096)/1024)` anchored at nbt 4096, so rungs
+1024/2048/4096 are plan-identical (the fitted 5.25 GiB was fitted AT 4096).
+MNBT changes re-key the compile range (one recompile, ~24 s at (1,2048) with
+the mounted cache). 512 doubles TTFT at 4064 ctx; 12288 costs ~2.5 % plus
+arena. MNBT does not move the KV pool.
+
+## 4. Registry "compile headroom" is not reclaimable slack
+
+`--expert-precise-memory` selects the fitted overhead (5.25 GiB, this arch at
+TP=2) over the default (7.60 GiB = fitted + **2.35 GiB of compile and
+graph-capture headroom** the registry documents as "transient VRAM the
+profiler never sees"). It moves exactly one term — the plan's `room` — so the
+arena grows by that 2.35 GiB (~1,025 slots). **That headroom is the budget
+for the phase that allocates LAST**: reclaiming it requires paying elsewhere
+(smaller pinned pool, fewer sessions, shorter context). Measured 2026-10-01:
+bf16 2×262,144 + precise put the total at 31.73 of 31.86 GiB and the boot
+crash-looped at graph capture; dropping the flag restored ~2.4 GiB and booted.
+
+## 5. Boot ≠ survival (knife-edge shapes)
+
+A shape can pass EVERY boot gate — weights, profile, KV allocation, graph
+capture, `/health` 200 — and die on the first real request. Measured
+2026-10-01: `5 × 262144 @ fp8` booted (KV 1,350,182 tokens / 5.15x) and was
+killed seconds later by a 995-token prompt with `max_tokens=261149`:
+`torch.OutOfMemoryError: Tried to allocate 50.00 MiB … 0 bytes free`. After
+any knife-edge change: send a LARGE-`max_tokens` request before calling it
+done, and read post-boot free VRAM per rank (`rocm-smi --showmeminfo vram`)
+— the survivor had ~3.6 GiB free; the corpse had 0.
+
+**KV accounting — pin the pool BYTES, floor-sized:** set
+`--kv-cache-memory-bytes` PER RANK to the contract floor
+(`seqs × MAXLEN × measured B/token`), not the native window — extra pool bytes
+come out of the arena/headroom and buy nothing. Measured B/token on this
+hybrid: **7,701 fp8 / 13,839 bf16** (decomposed: attention 6,138/12,276 +
+dtype-independent mamba/side 1,563). Measured rates are shape-dependent —
+7,891.8 at the 200704×3 shape. Capacity does NOT scale with `SEQS`: the
+planner's KV reserve is for ONE max-model-len request.
+
+**Parallelism ceilings you cannot flag past:**
+
+- `--decode-context-parallel-size` refused for GQA when `TP ≤ num_kv_heads`:
+  *"requires TP (2) to be greater than the model's total number of KV heads
+  (2)"* — DCP shards KV along the sequence dimension AFTER KV heads spread
+  across TP ranks (it reuses TP ranks, not TP×DCP), so a 2-KV-head model
+  needs TP ≥ 4 = four GPUs.
+- EP (`--enable-expert-parallel` + `--all2all-backend`) exists but REPLACES
+  the fork's local tiered-offload gather with a per-layer PCIe all-to-all.
+- DP requires a full model copy per replica — on a model larger than the
+  card it is impossible, not slow.
+
+## 6. The context ladder (find the max bootable window by measurement)
+
+"Max possible" = the largest value that BOOTS, measured — never the model's
+spec maximum, never hand arithmetic.
+
+1. **Scale the pool with the window.** If KV_BYTES is pinned, it must shrink
+   in lockstep with MAXLEN or the rung tests nothing:
+   `KV_BYTES = round(KV_BYTES_ref × (seqs×MAXLEN_new) / (seqs×MAXLEN_ref))`
+   off a booted reference rung. Verify the rung's capacity still covers
+   `seqs × MAXLEN` from the engine's own line.
+2. **Boot under the PRODUCTION environment, not a shell probe** — a rootless
+   shell lacks `LimitMEMLOCK=infinity` and can pass where production OOMs.
+   Transient unit:
+   `sudo systemd-run --uid=<svcuser> --unit=probe-$ML -p LimitMEMLOCK=infinity -p Restart=no env <ENVFILE_OVERRIDE> <serve-script>`
+   pointing the serve script at a scratch env file via its `*_ENV_FILE` hook.
+3. **Judge on health + unit liveness:** poll `/health` (200 up / 503 loading
+   / 000 not bound); break early when the unit goes inactive — that is the
+   failure signal, not a timeout.
+4. **Capture the engine log per rung BEFORE teardown** (`podman logs` dies
+   with `podman rm`; a `podman logs` issued too soon after `podman run -d`
+   fails with *no container with that name* — record UNKNOWN, never FAIL).
+5. **Ladder down on FAIL, one bonus step UP on PASS.** Measured steps on this
+   box: OOM −16,384 tokens, bonus +4,096. Tiny margins (76 MiB class) mean
+   the next rung down almost certainly passes: `Tried to allocate 76.00 MiB …
+   0 bytes free … 30.43 GiB allocated by PyTorch` died in graph CAPTION;
+   `Tried to allocate 10.28 GiB … 9.46 GiB free` died in KV ALLOCATION.
+6. **Clean between rungs:** stop the transient unit, `podman rm -f`, kill
+   orphans by explicit PID
+   (`ps aux --sort=-rss | awk '/vllm\/bin\/python3/ && $6 > 50000000 {print $2}'`),
+   wait for `MemAvailable` to recover.
+
+**The hazard that cost production:** a leftover ladder script whose teardown
+`podman rm -f`s the production container fires against a freshly DEPLOYED
+engine. Teardown runs ONLY on failure, never after a pass; kill the whole
+scratch process tree and verify none remain BEFORE deploying. On PASS: print
+the measured `GPU KV cache size / Maximum concurrency` line, LEAVE IT
+SERVING, and pin the number into the repo (env + seam + harness
+contextWindow) before deploying.
+
+Cutover order: ladder → winner passes and serves → repo env + seam + harness
+→ tests → commit+push → verify no scratch scripts remain → stash foreign work
+(`git stash -u`) if the tree is dirty → deploy (one restart) → verify health
++ engine flags in the log.
+
+**Measured ladder on this box (200704×3, fp8 KV, precise memory):**
+
+| shape | util | graphs | arena | result |
+|---|---|---|---|---|
+| E | 0.98 | 0.55 | 17.22 GiB / 7,522 slots | boot ok |
+| F | 0.982 | 0.32 | 17.52 GiB / 7,653 slots | boot ok |
+| 1.0 | 1.00 | 0.07 | — | UNBOOTABLE (worker-init ValueError, both ranks) |
+| G (deployed) | 0.982 | 0.07 | 17.77 GiB / 7,762 slots | KV 401,408 tokens / 2.00x; 0.89 GiB real free/card held through a live 4,096-token generation at ~38–39 tok/s |
+
+bf16 ceilings at util 0.98: 3 seqs → 241,664 passes / 245,760 and 262,144 OOM
+at graph capture; 6 seqs fp8 → 10.70 GiB ask with 9.18 GiB free (fails) —
+~207k per context max.
+
+## 7. Arena audit procedure (when asked for "more experts")
+
+1. Read the live plan before proposing anything — `grep -oE` with a
+   digit-anchored tail (plain grep/journalctl truncate these columns to
+   `17.`):
+
+   ```
+   ssh dankserver 'sudo journalctl CONTAINER_NAME=frank-primary-qwen -b -o cat --no-pager \
+     | grep -aoE "(budget \(total x util\) +[0-9.]+|expert cache space +[0-9.]+ GiB +[0-9,]+ slots|Estimated CUDA graph memory.{0,80}|Graph capturing finished.{0,40}|GPU KV cache size.{0,90})"'
+   ```
+
+2. Classify each term (§1 table); take only the two reachable levers.
+3. Verify after deploy: plan lines show new budget/graphs/arena;
+   `GPU KV cache size` matches the ask; graph-estimate vs post-capture;
+   `rocm-smi` high-water per card = the real margin;
+   `systemctl show frank-primary -p NRestarts --value` = 0; a real chat
+   completion from a PEER (health-probe via `192.168.0.235:7770/health`, not
+   `localhost` inside an ssh session on the host — the latter returns `000`
+   and looks like an outage).
+
+## Related
+
+- [[RDNA4 Container Serving]] — the bring-up path and container traps
+- [[vLLM Boot Forensics and Config Interrogation]] — capture and contracts
+- [[dankserver-qwen38-fp8-expert-offload-tuning]] — measured history
+- [[Llama Cpp Serving Sizing]] — the llama.cpp equivalent

serverAppConfigs-docs/LLM/Tcclaviger Fork Ops.md

@@ -0,0 +1,264 @@
+---
+type: reference
+title: Tcclaviger Fork Ops
+description: Operating the tcclaviger RDNA4 vLLM fork on dankserver (shape G era, rollback target since the 2026-10-05 FreeToken cutover) — the ad-hoc sweep harness, measured knob verdicts, sweep hygiene, the large-MoE serving recipe, and compile-cache persistence.
+tags: [tcclaviger, vllm, dankserver, expert-offload, sweeps, cache]
+timestamp: 2026-10-05T00:00:00-07:00
+---
+
+# Tcclaviger Fork Ops
+
+The tcclaviger RDNA4 vLLM fork (`docker.io/tcclaviger/vllm@sha256:da4029666…`,
+vLLM 0.29.0.dev, GLM-5.3/Qwen-specialised with native MoE quant kernels and a
+first-class expert-offload planner) served `frank-primary` at shape G until
+the 2026-10-05 FreeToken cutover; it remains the rollback target
+(`git tag frank-primary-vllm-shapeG-20261003`). Full plan law and VRAM
+arithmetic: [[RDNA4 MoE Expert Offload VRAM Budget]]; boot forensics:
+[[vLLM Boot Forensics and Config Interrogation]]; window discipline:
+[[LLM Bench Windows and Measurement]].
+
+## The ad-hoc sweep harness (dankserver)
+
+Serve script `/home/dank/serve_qwen_fp8.sh` with env hooks: `SPEC` (MTP
+depth), `TUNABLE` (TunableOp), `PREFETCH`, `UTIL`, `MAXLEN`, `KV_BYTES`,
+`MNBT`, `UVA_TOK`, `GATHER_C/GATHER_L` (R4D_LRU gather), `TELEM`
+(R4D_LRU_TELEMETRY), `PROFDIR` (torch profiler bind-mount + --profiler-config
+JSON). The script re-copies the harness (`mtp_measure.py`, `bench_iter.py`)
+into every fresh container — `podman cp` files die on container recreate.
+
+- Measure ONLY via usage counters: `mtp_measure.py` (3 depths) or
+  `mtp_measure_short.py` (166 ctx). `decode = completion_tokens/(wall −
+  ttft)`; chunk counting is INVALID under spec decode.
+- Jobs: `bash /home/dank/job.sh start NAME CMD…` / `wait NAME SECS 40` /
+  `out NAME` ([[Remote Job Runner]]).
+- **Validity check on ANY speed result:** `completion_tokens == max_tokens`
+  (else repetition-halt or early EOS), finish_reason, and inspect the text —
+  `bench_iter` prompts are repeated word-lists + `!` padding, loop-prone. A
+  loop makes LRU hits ~100% → PCIe traffic vanishes → fake 3–4× decode.
+  Acceptance swings 48–61% run-to-run — noise; decode is the metric.
+- Each arm = full reload: ~8.5 min cold, ~12 min with measurement.
+- Readiness gate: `podman logs <c> 2>&1` (stdout+stderr merged) for
+  `Application startup complete` — it is on STDERR; an unmerged grep can
+  never match and burns the whole wait timeout.
+- GPU util ~0% during load/compile is NORMAL (weights + CUDA-graph capture
+  are CPU-side); VRAM climbing is the load indicator.
+- Config edit discipline: the serve script is patched by scp'd python
+  scripts with asserts (anchor must exist, addition must not); `bash -n`
+  after every patch.
+
+## Measured knob verdicts (2026-09-29 campaign)
+
+Incumbent: `SPEC=4 PREFETCH=1 UTIL=0.98 MAXLEN=32768 KV_BYTES=2147483648
+TUNABLE=0`, stock gather 8×16 → 26.80/25.15/26.81 tok/s @ 166/1087/4064 ctx.
+
+| Knob | Verdict | Evidence |
+|---|---|---|
+| Gather geometry `R4D_LRU_GATHER_{CHUNKS,LANES}` | **RULED OUT** | 16×32 worse (~24); 32×64 tie (26.6). The 9.5 vs 22.7 GB/s read-path gap is per-transfer structure (~105 µs/slab vs 78 ideal), not a tunable knob |
+| MTP depth | **RULED OUT** | SPEC=3 ~9% worse; SPEC=6 collapses (13.5–19 tok/s, acceptance 17.4%) — also proves UVA read-through (crosses at step tokens > `R4D_LRU_UVA_TOKENS=5`) is SLOWER than copy-in |
+| TunableOp | **REJECTED** | `CLAV_TUNABLEOP_SWEEP=1` silently SKIPPED under expert offload unless `CLAV_TUNABLEOP_WITH_OFFLOAD=1`; adds ~7 min/load, persists nothing (container-local), and CORRUPTS NUMERICS (loops, repetition-halt, fake 79.8 tok/s @ 96.5% acceptance on degenerate completions). Any "3× win" with early-EOS + absurd acceptance = the loop artifact |
+| MNBT | neutral at 8192 vs 2048 | small values re-chunk prefill only (hurts TTFT, never decode — decode step = max-num-seqs × (spec+1) tokens). No ubatch concept in vLLM |
+
+Decode on this stack is PCIe-gather-bound (~30% GPU duty cycle), so
+batch-shaped compute knobs move ~nothing.
+
+The `R4D_*/CLAV_*` knob family (QSA_*, MHC, ROUTE, ACT_QUANT, AR_QUANT,
+SILU_QUANT) are kill-switches defaulting ON, not perf dials; real levers are
+the R4D_LRU_* family, TunableOp (rejected), R4D_LRU_TELEMETRY, and spec
+depth.
+
+Telemetry gotchas: `--profiler-config` takes a JSON blob, not separate
+flags; `torch_profiler_with_stack` defaults TRUE — a full measure inside a
+profile window OOM-kills the engine worker (~40 GB trace buffer; kernel
+kills python3 at ~95 GB anon-rss). Keep windows to ~15 s / one depth
+(`mtp_measure_short.py`). `telemetry_report` (`expert_cache_rank<N>.json`)
+fires ONLY from the worker's profiler-stop path; worker-side stop can hang
+>5 min on ROCm/kineto. No alternative trigger exists. The LRU hit rate
+measured 82.45% — NOT the ~27% the arena-fraction arithmetic implies;
+temporal reuse dominates, so never infer miss rates from resident-slot
+fractions.
+
+## Sweep hygiene (fork-specific preflight)
+
+1. **Knob landed:** for each `-e` env knob, confirm in-container
+   `podman exec … env` (or the engine log's confirmed-values line). A knob
+   the fork silently ignores produces a "parity" arm that is really the
+   default — the v1 gather sweep lost all three arms to exactly this
+   (GATHER_C/GATHER_L silently dropped, all arms ran stock 8×16).
+2. **Prefetch uniformity:** every arm in a sweep MUST carry the same
+   expert-prefetch setting. The 2026-09-29 campaign measured losing arms
+   with prefetch ON and incumbents OFF; untaxed, its verdicts inverted
+   (no-spec 12.11→41.47; gather 32×64 parity→+18.7%). Never re-read a taxed
+   verdict.
+3. **Spec arms must prove engagement:** engine log
+   `SpecDecoding metrics: Mean acceptance length …` and
+   `curl :7770/metrics | grep vllm:spec_decode_`. Zero such lines = MTP off.
+4. Boot-fail fast-detection: poll `podman ps` INSIDE the health loop; a
+   refused boot dies in one poll instead of 20 min of health waits. Capture
+   >8 lines of `podman logs` on failure — the root cause (e.g. a SEQS init
+   refusal) lives higher up.
+5. Report per-depth tables, never one number: g32×64's win grows with depth
+   (54@1k, 39.5@short) — a single 4064 median hides the short-ctx cost. TTFT
+   rows are noisy ±0.15 s and non-monotonic across depths; compare same-ctx
+   rows across arms only.
+6. Generate ≥256 tokens per sample (at 8 tokens inter_token is first-token
+   noise).
+
+Boot-log facts survive measure failures: KV pool size (`GPU KV cache size`),
+arena slots (`expert cache: … -> N slots`), and the spec-reservation ladder
+(MTP3 766,666 / MTP4 729,545 / MTP6 665,263 tokens) print at BOOT — a
+measure-stage bug costs decode tables but not boot-log findings.
+
+**MTP depth-brittleness law (2026-10-02, all arms rc=0):** MTP3 on this fork
+is depth-brittle — 47.01 t/s at 166 tokens (47.88 at temp 0.7) but 23.09 at
+4064, vs no-spec baseline 37.02 / 33.99. MTP3 + SEQS 16 + MNBT 4096 + util
+0.95 + ladder [4..32] = best overall 37.04 / 38.61 t0.7 and keeps 33.15 at
+4064. `--expert-precise-memory` is load-bearing for MTP (same arm with it
+OFF = worst arm, 19.81 at 4064). Every spec arm costs KV (398,658 → 313,002
+tokens). Acceptance is real: mean length 2.4–2.8, per-position
+0.76/0.55/0.33, avg 52–57%. Next lever named by the engine: MNBT 4096 is
+exactly SEQS×(n+1) (`suboptimal` warning) → try 8192.
+
+## Serving a model larger than VRAM (dankserver recipe)
+
+Dankserver: 2× R9700 (gfx1201), 31.86 GiB each = 63.7 GiB real VRAM, 246 GiB
+RAM. `torch.cuda.device_count()` returns 3 and summing `total_memory` gives
+a bogus 186.7 GiB (the Raphael iGPU is in the list — see
+[[Llama Cpp Serving Sizing]] for the reading discipline).
+
+**Kernel gate — this decides which checkpoint can work at all.**
+`/app/r4dhip/r4d.so` exports `gemm_moe_mxfp4a8_*` and `gemm_moe_dsfp4a8_*`
+but NO `gemm_moe_w4a16_*`. Expert offload requires an experts class with
+BOTH a compiled kernel AND an `offload_recipe`; the planner aborts startup
+otherwise. Usable: MXFP4, DSFP4, FP8-HIP (`R4dMxfp4MoEExperts`,
+`R4dDsfp4MoEExperts`, `Fp8HipMoEExperts`). `R4dW4a16MoEExperts` has no
+kernel (the class exists, the r4d.so entry does not); RFA has neither. Check
+the checkpoint's quant against this list BEFORE downloading anything.
+
+Entrypoint is `/app/tools/image_entrypoint.sh`, ending `exec vllm serve
+"$@"` — pass the model path as the FIRST argument and do NOT pass `serve`.
+Known-good invocation:
+
+```bash
+sudo sh -c 'sync; echo 3 > /proc/sys/vm/drop_caches'
+podman run -d --name <ctr> --ipc=host --ulimit memlock=-1:-1 \
+  --cap-add SYS_PTRACE \
+  --device /dev/kfd --device /dev/dri --device /dev/mem \
+  -e HIP_VISIBLE_DEVICES=0,1 -e VLLM_ROCM_USE_AITER=0 \
+  -v /home/dank/models/<model>:/models:ro \
+  docker.io/tcclaviger/vllm:latest /models \
+  --served-model-name <name> --port 7770 \
+  --tensor-parallel-size 2 \
+  --max-model-len 32768 \
+  --kv-cache-memory-bytes 2147483648 \
+  --max-num-seqs 4 \
+  --gpu-memory-utilization 0.95 \
+  --enable-prefix-caching --trust-remote-code
+```
+
+`-p 7770:7770` is not included, so the API is only reachable via
+`podman exec` — add `-p` if host-side curl is needed. Load ~5 min;
+`/health` answers connection-refused the whole time; watch for
+`Application startup complete` in `podman logs <ctr>`. Verified working:
+`orcarouter/Qwen3.8-Flash-Next-Uncensored-FP8` (gated; 172.8 GiB, 131
+shards, `Qwen4ExpForConditionalGeneration`, 48 layers, 512 experts top-10,
+multimodal, native 262,144).
+
+Traps that each cost a boot:
+
+1. **Mount the whole `/dev/dri`, never individual render nodes.** Render
+   nodes without the primary `card*` nodes → HIP enumerates two agents and
+   refuses a context on either (`RuntimeError: No CUDA GPUs are available`)
+   while `device_count()` cheerfully returns 2. Confirmed by direct A/B.
+2. **memlock must be raised** (done on Dankserver via
+   `/etc/security/limits.d/99-memlock.conf`). Host default 8 MiB makes any
+   large pinned allocation fail as a misleading
+   `torch.AcceleratorError: CUDA error: out of memory` from `pin_memory()`;
+   it cannot be raised from inside a container or root's own shell — only
+   PAM applies it at login. Matters because `--expert-host-mlock` defaults
+   True.
+3. **`--gpu-memory-utilization` does NOT buy KV under expert offload** —
+   the planner sizes the arena against the budget and leaves KV the
+   remainder; raising util 0.95→0.98 grew the arena and the KV pool stayed
+   at 0.64 GiB.
+4. **Lowering `--max-model-len` is the WRONG lever and backfires** — the
+   budget subtracts the KV reserve first, so a smaller context shrinks the
+   reserve and hands the space to the arena (measured: 65536 → 0.64 GiB KV;
+   32768 → 0.34 GiB).
+5. **The KV reserve undercounts hybrid-attention models ~2×** (9190 B/token
+   planned vs ~18,000 real: the Mamba state is a fixed per-sequence cost).
+   Fix by pinning the pool: `--kv-cache-memory-bytes`.
+6. **`--expert-offload-mem` caps the HOST side**, not the VRAM arena —
+   structurally orthogonal to `--expert-precise-memory` (verified
+   in-container 2026-10-05: precise-memory only swaps the GPU overhead term,
+   expert_offload_mem only sets the host bound; they rejoin only inside
+   `place_experts`, and the plan rewrites `expert_offload_mem` back to the
+   cap the plan met, so plan and allocation cannot drift).
+7. `--ipc=host` and `--shm-size` are mutually exclusive in podman; pick one.
+8. `podman run -d` without `-a` sends no stdout to a redirect — the log file
+   only receives the container ID; use `podman logs`.
+
+Status reads: use `scripts/dankstatus.sh`, never hand-rolled
+`/proc/meminfo` parsing — `/proc/meminfo` separates key from value with a
+TAB, so `grep MemAvailable | tr -s " " | cut -d" " -f2` returns empty and
+awk prints `0.0` — a false "host out of memory" that invites an unnecessary
+abort. Gated checkpoints: token via a mode-600 `curl -K` config, never argv
+(`pgrep -a curl` prints it); always `--fail`; compare each file's byte count
+to the Hub's declared size; `-C -` resumes truncated shards.
+
+**Check whether the stack supports the model natively before porting an
+overlay.** The image already registered `Glm5NextForConditionalGeneration`,
+`Qwen4ExpForCausalLM` and both MTP variants — chasing four missing symbols
+through two image tags to bring an overlay up was entirely wasted work.
+
+## Compile-cache persistence (survives `podman rm -f`)
+
+Serve scripts that recreate the container discard the writable layer, so
+every boot re-pays compile/JIT (frank-primary: `torch.compile took 181.20 s`
+cold, of which 173 s was one graph).
+
+1. **Prove it and locate it:** fresh container vs live
+   (`podman run --rm --entrypoint bash IMG -c "ls -la /root/.cache"` vs
+   `podman exec LIVE bash -c "du -sb /root/.cache/*; find /root/.cache -type
+   f | wc -l"`). vLLM cache root: `VLLM_CACHE_ROOT` (default
+   `/root/.cache/vllm`, `torch_compile_cache/`); a HIP runtime adds
+   `/root/.cache/comgr`.
+2. **Mount the PARENT, not the leaf** — one bind on `/root/.cache` persists
+   both trees, safe whenever the image ships no such directory (step 1
+   proves it; otherwise you would be hiding baked state). Wire as an opt-in
+   tunable: `mkdir -p "$CACHE_DIR"` BEFORE the mount, then
+   `${CACHE_DIR:+-v "$CACHE_DIR":/root/.cache}`. Host path outside the
+   checkout.
+3. **Verify with two consecutive boots** — cold `Compiling a graph … takes
+   N s`, warm `Directly load the compiled graph(s) for compile range (…)
+   from the cache` with `torch.compile took` in single-digit seconds.
+4. Traps: the cache key includes the compile RANGE — an MNBT change costs
+   ONE recompile, not a permanent tax; do not claim a proportional boot cut
+   (time to `/health` is dominated by weight loading: 490 s warm vs 188 s
+   cold compile); stale entries are misses, not wrong answers (version +
+   config hashed); if the serve script is seam-gated, assert the mkdir and
+   the exact mount target.
+
+## Plan-line verification checklist (after any fork deploy)
+
+```
+ssh dankserver 'sudo journalctl CONTAINER_NAME=frank-primary-qwen -b -o cat --no-pager \
+  | grep -aoE "(budget \(total x util\) +[0-9.]+|expert cache space +[0-9.]+ GiB +[0-9,]+ slots|Estimated CUDA graph memory.{0,80}|Graph capturing finished.{0,40}|GPU KV cache size.{0,90})"'
+```
+
+`grep -oE` with a digit-anchored tail — plain `grep`/journalctl truncate
+these columns to `17.`. Verify: new budget/graphs/arena in the plan;
+`GPU KV cache size` matches the requested token count; `Estimated CUDA graph
+memory` vs the capture line (raise `_PER_GRAPH_MB` toward 52 if capture
+exceeds the charge); `rocm-smi --showmeminfo vram` high-water per card = the
+real margin; `systemctl show frank-primary -p NRestarts --value` = 0;
+`VLLM_CUDAGRAPH_*` present in `podman exec … env`; then a real chat
+completion from a PEER — `curl localhost:7770/health` inside an ssh session
+on the host can return `000` and looks like an outage that isn't.
+
+## Related
+
+- [[RDNA4 MoE Expert Offload VRAM Budget]] — the plan law this harness tunes
+- [[vLLM Boot Forensics and Config Interrogation]] — journal forensics for these boots
+- [[LLM Bench Windows and Measurement]] — window discipline and bash laws
+- [[Model Swap and Rollback]] — where the fork sits in the swap history
+- [[Remote Job Runner]] — the job.sh harness referenced above

serverAppConfigs-docs/LLM/VLLM Overview.md

@@ -237,6 +237,46 @@ vLLM supports multiple parallelism strategies:
 - Expert Parallelism (EP) distributes expert networks in MoE models
 - Context Parallelism (CP) splits context across GPUs
 
+
+## Chunked Prefill — the flag surface, and what it actually does
+
+**YES, the runtime exposes chunked-prefill options, and the running shape
+already uses them.** The tcclaviger fork is a vLLM fork, so the standard
+scheduler flags are present in the shipped image (verified by reading the
+`vllm.engine.arg_utils` and `vllm.config.scheduler` source in the live
+container `frank-primary-qwen`, not from a stub):
+
+| Flag | Default | What it controls |
+|---|---|---|
+| `--enable-chunked-prefill` | `True` in `SchedulerConfig`; the CLI arg is an auto-flag (`"default": None`) | whether prefill may be split into chunks |
+| `--max-num-batched-tokens` (MNBT) | 2048 (`DEFAULT_MAX_NUM_BATCHED_TOKENS`) | tokens processed per prefill iteration |
+| `--prefill-schedule-interval` | 0 | gap between prefill iterations |
+
+The running shape's argv carries `--max-num-batched-tokens 2048` alongside
+`--max-model-len 200704`, so chunked prefill is **ON and already active**: a
+200704-token prompt is prefilled in ~98 iterations of 2048 tokens, not in one
+shot.
+
+**It is a TTFT lever, not a throughput lever.** MEASURED 2026-09-29 on the
+ad hoc qwen-fp8 container (the same fork, same payload, same TP=2):
+
+- MNBT 8192 vs the default 2048: **neutral** on decode at every depth.
+- Small values re-chunk prefill only. They hurt TTFT (each chunk is its own
+  iteration, so the first token waits on more iterations) and they never touch
+  decode: a decode step is `max_num_seqs x (spec+1)` tokens, independent of
+  MNBT.
+
+**There is no `--ubatch-size` equivalent in vLLM.** `--ubatch-size` is a
+llama.cpp compute-buffer concept (`--batch-size`/`--ubatch-size`); the tcclaviger
+fork has no ubatch knob at all. The llama.cpp compute-buffer pair that frank-small
+was doubled on (2026-10-05, +4.6 % prefill at 8,928 tokens for 616 MiB) has no
+counterpart here, and the two backends should not be reasoned about by analogy.
+
+**MNBT is not a lever on this shape either.** The deployed 2048 is the fork's
+default, so a change is a deliberate departure from the shipped value, not a
+tune. The one thing MNBT does cost: every change re-enters the compile RANGE
+path, so `FRANK_PRIMARY_CACHE_DIR` pays one recompilation.
+
 ## See Also
 
 - [[VLLM Index]]

serverAppConfigs-docs/LLM/dankserver-frontier-moe-feasibility.md

@@ -0,0 +1,227 @@
+---
+type: research
+title: Frontier MoE on Dankserver — what fits, on which runtime
+description: VERIFIED 2026-09-28 by measurement and source reading. Compares DeepSeek-V4.1-Flash, MiMo-V2.6-Flash-MOPD and GLM-5.3-Flash against the 2x R9700 (gfx1201) / 246 GiB box, and assesses the Affinity runtime's portability off DeepSeek-V4-Flash. Records the capacity arithmetic that decides the answer before any model is loaded.
+tags:
+  - dankserver
+  - r9700
+  - gfx1201
+  - moe
+  - deepseek-v4.1
+  - deepseek-v4
+  - mimo
+  - glm
+  - affinity
+  - lucebox
+  - feasibility
+timestamp: 2026-09-28
+---
+
+# Frontier MoE on Dankserver — what fits, on which runtime
+
+> **Question this page answers.** Given Dankserver's hardware, which of
+> DeepSeek-V4.1-Flash, MiMo-V2.6-Flash-MOPD and GLM-5.3-Flash runs best, on
+> which runtime, and can Affinity / vllm-radiance / FreeToken-AMD be modified
+> to run them fast?
+>
+> **Answer in one line.** V4.1-Flash does not fit this box at any
+> configuration; the engine that does run it (Lucebox) measures
+> 0.8–9.1 tok/s on R9700 because decode is bound by PCIe expert streaming
+> rather than by the GPUs. The 284B **V4-Flash** on **Affinity** is the only
+> shape that clears 35 tok/s, at 110.
+
+Everything below was measured or read from source on 2026-09-28. Claims
+carried over from the engine authors are marked as theirs.
+
+## Hardware (measured live, 2026-09-28)
+
+| Property | Value |
+|---|---|
+| GPUs | 2x AMD Radeon AI PRO R9700 (Navi 48, **gfx1201**), 32,624 MiB each = **63.7 GiB** total |
+| Interconnect | PCIe 5.0 **x8** each, CPU root ports `00:01.1` / `00:01.3`, ~31.5 GB/s per card |
+| PCIe atomics | `CAP_EXP+28.w = 0x1440` (ReqEn+) on both endpoints — RCCL collectives are legal |
+| CPU | Ryzen 7 7700X — **8 cores / 16 threads** (small; CPU expert compute is expensive) |
+| RAM | **246 GiB** + 63 GiB swap |
+| Disk | 1.8 TB NVMe. `/mnt/entropy` is a separate 91 TB ext4 at **100%** — do not plan payload there |
+
+The knowledge base was wrong about RAM: [[Dankserver System Specs]] said
+60 GiB until 2026-09-28 and has been corrected. The Lucebox runbook's 246 GiB
+was right.
+
+## The three candidates
+
+| Model | Total / active | Routed experts | Attention & KV | Working gfx1201 runtime |
+|---|---|---|---|---|
+| **DeepSeek-V4.1-Flash** | 552B / 8B prefill, 16B decode | 384 top-6, 40 layers, `moe_intermediate_size` 2304, hidden 5120 | CED 20 enc + 20 dec; CSA2 Full/Reindex/Reuse; **FP4 main KV, 890 B/token**; Engram 196B at layers 1 & 14 | Lucebox `feat/ds41-engine` only (unmerged) |
+| **MiMo-V2.6-Flash-MOPD** | 309B / 15B | 256 top-8, no shared expert, 48 layers (39 SWA / 9 GA) | hybrid SWA (window 128) + GA; 5-layer MTP | vLLM / SGLang, NVIDIA-first. No AMD path |
+| **GLM-5.3-Flash** | 320B / 18B | MoE | KDA (linear) + DSA (sparse) hybrid; mHC | **tcclaviger vLLM fork — plausible native path (2026-09-29 source reading: `glm5next` registered with an `amd/` backend dir + DSA indexer + mHC path; r4d.so carries native mxfp4/dsfp4 MoE entries). Gate: a compressed-tensors MXFP4 payload (~165 GiB, none exists publicly) + a HIP boot probe.** llama.cpp GGUF (community R9700 guide) remains the verified-but-slower path |
+
+## Sizing — the arithmetic that decides it
+
+| Quantity | Value | Note |
+|---|---:|---|
+| DS4.1 official checkpoint | **475.27 GiB** | 48 shards, summed from the HF API |
+|   of which Engram | **189.13 GiB** | exactly two 94.56 GiB shards — the layer 1 and 14 lookup tables |
+| DS4.1 routed-expert params | 543.6 B | 40 x 384 x 3 x 5120 x 2304 |
+|   shipped as | 269 GiB (fp4) | DeepSeek's own `expert_dtype: fp4` |
+| Lucebox `ROCMFP23.gguf` | **377.51 GiB** | 179 GiB experts (2.5/3.5 bpw) + 189 GiB Engram + ~9.5 GiB dense |
+| V4-Flash official checkpoint | 148.67 GiB | **3.2x smaller than V4.1** |
+| Affinity container for V4-Flash | 101.1 GiB | 92.9 GiB experts @ 2.875 bpw + 6.2 dense + 2.0 embed |
+
+The Engram tables are **never mapped** by Lucebox, so they do not evict the
+expert page cache — that is the one piece of good news in the 377 GiB file.
+
+### The decode wall
+
+A V4.1 token selects `6 x 40 = 240` routed experts at ~11 MB each. At the
+residency the hot stack can hold (37 of 384 per layer, 9.6%), **~2.4 GB of
+expert weights cross PCIe per token**. PCIe 5.0 x8 does ~55 GB/s real:
+
+```
+2.4 GB / 55 GB/s = 44 ms  ->  ~23 tok/s absolute ceiling
+```
+
+That is **below the 35 tok/s target before any attention, Engram or dense work
+is counted.** Raising residency is the only lever, and VRAM is what bounds it:
+63.7 GiB / 11 MB = 5,800 of 15,360 experts = 37.8% maximum, and that assumes
+VRAM holds nothing else.
+
+The engine author measured the same wall on an R9700 with 61 GB of RAM
+(DS41.md, "Measured"): 0.8 tok/s cold / 1.8 tok/s warm at a 6 GB hot budget,
+0.9 / 1.9 at 16 GB, and 9.1 tok/s for the best tuned three-tier case. They
+name the cause: *"90 to 96% of the 240 routed experts a token selects are still
+cold, and decode moves roughly 2.4 GB of expert weights per token through the
+synchronous stream path. That, not the GPU, bounds decode today."*
+
+## Affinity — a DeepSeek-V4-Flash engine, and what a V4.1 port costs
+
+[Affinity](https://codeberg.org/StillDeadcode/affinity) is *"an inference
+engine for DeepSeek-V4-Flash on one or two RDNA4 GPUs"*, C++/HIP, no framework.
+**It builds clean on this host**: CMake 3.24+, `hipcc` 23, `HIP enabled:
+gfx1201`, zero errors, producing `affinity`, `aff-quantize`, `aff-info` and
+`aff-gpucheck`. Its README claims **~110 tok/s on two R9700s** against 8 on
+one. That is 3x Frank's 35 tok/s goal, for the 284B model.
+
+The 110 works because the 101 GiB container leaves the cold expert set in a
+**pinned host pool in RAM** — the SSD tier that Affinity calls "a cliff" is
+never reached. V4.1 would need 182 GiB of experts at the same 2.875 bpw, plus
+189 GiB of Engram, against the same 246 GiB of RAM.
+
+### What a V4.1 port would reuse
+
+The config parser is **fully data-driven** — every dimension is read from
+`config.json`, so nothing is compiled in at the config layer. And the expert
+tensor naming is *identical* between the two checkpoints:
+
+```
+layers.N.ffn.experts.E.w1.weight   (11,008 in V4  = 43 x 256)
+layers.N.ffn.experts.E.w1.weight   (15,360 in V4.1 = 40 x 384)
+```
+
+So the quantiser, the `.aff` container, the expert slab, the host pool and the
+heat/placement engine — **the part that produces the 110 tok/s** — all carry
+over. Hyper-connections are the same family (`hc_mult 4`, Sinkhorn 20).
+
+### What a V4.1 port would have to write
+
+Verified against both checkpoints' real `model.safetensors.index.json` on
+2026-09-28 — not inferred from the config prose.
+
+**Tensor presence.** `aff-quantize` requires 32 per-layer tensors. V4.1
+supplies 26 plus every global (`embed`, `head`, `norm`, `hc_head_*`) and the
+whole expert set. Six are missing:
+
+| Missing in V4.1 | Replaced in V4.1 by |
+|---|---|
+| `ffn.gate.tid2eid` | the Engram tables |
+| `attn.compressor.ape` | nothing — ratio-2 is *pooled, no APE* |
+| `attn.indexer.compressor.{wkv,wgate,ape,norm}.weight` | `attn.indexer.wk` + `attn.indexer.k_norm` |
+
+**Config keys.** Of the 35 Affinity reads, V4.1 supplies **34**. The one it
+lacks is the dangerous one:
+
+| Gap | Detail |
+|---|---|
+| **Engram — fails silently** | `num_hash_layers` (V4: 3) is absent, so `cfg_int` returns its default and `model.h:36`'s `n_hash_layer = 3` survives. It would route layers 0–2 through a table that does not exist: fluent output, wrong numbers, no error. This is the only silent failure; the other four gaps fail loudly. |
+| **Ratio vocabulary is disjoint** | V4 uses `{0, 4, 128}`; V4.1 uses `{0, 1, 2}` — **no overlap in the non-zero case**. Affinity computes `coff = (ratio == 4) ? 2 : 1`, so V4.1's 18 ratio-2 layers get the ratio-1 width; `comp_ape` is requested 38 times for a tensor that does not exist; and the `ratio == 4` indexer branch fires on **0** of the 8 layers that need one (`index_source_layer_ids` = 2, 8, 14, 20, 24, 28, 32, 36) — those 8 would run dense. |
+| **CED** | A 20-layer encoder feeding a 20-layer decoder whose global KV is projected from the final encoder hidden state, through an `aligner`. Affinity's `model.cpp` is a flat layer stack. |
+| **DSpark** | V4.1's draft carries **its own** 128-expert MoE at top-3 (`dspark_n_routed_experts`, `dspark_num_experts_per_tok`), 3 stages rather than 1. Affinity's draft shares the target's slab and routing. |
+| Vision | V4.1 is multimodal; Affinity has no vision path. Text-only is an acceptable loss. |
+
+V4.1 also adds 31 config keys Affinity never reads, including the whole
+`engram_*` block, `kv_source_layer_ids`, `index_source_layer_ids` and the
+candidate block pre-selection.
+
+**Verdict.** The reusable half is the *right* half — the expert quantiser,
+the `.aff` container, the slab, the host pool and the heat/placement engine
+are what produce the 110 tok/s. The rest is 6 tensor bindings, one new ratio
+vocabulary, the CED aligner, a new draft geometry and Engram: weeks of work
+in a third-party codebase, with no cheap way to prove correctness. And it
+meets the same capacity wall regardless, because 182 GiB of experts plus
+189 GiB of Engram against 246 GiB of RAM is exactly the problem Lucebox has.
+Not advisable.
+
+## The other runtimes
+
+| Runtime | gfx1201 | Architectures | Note |
+|---|---|---|---|
+| **Affinity** | yes, built here | **V4-Flash only** — "the geometry is compiled in; there is no architecture abstraction" | 110 tok/s for 284B. Best engine for this box by an order of magnitude, wrong model |
+| **tcclaviger vLLM fork** | **yes — proven 2026-09-29** | qwen4exp measured 37.4-40.6 tok/s decode (see [[dankserver-qwen38-fp8-expert-offload-tuning]]); **`glm5next` registered** (`Glm5NextForCausalLM`/`MTP`, backend-split `amd/`/`common/`/`nvidia/`, `sparse_indexer.py` for DSA); r4d.so has **native mxfp4 + dsfp4 MoE entries** (`has_dsfp4() == True`); W4A16 MoE class (`R4dW4a16MoEExperts`) wired into the same tiered offload; TunableOp silently skipped under offload (needs `CLAV_TUNABLEOP_WITH_OFFLOAD=1`, and then corrupts numerics — do not use) | the ad hoc qwen-fp8 container's runtime; **the old "MXFP4 = emulation/missing kernels" verdict predates this fork and is dead** — the GLM-5.3-Flash gate is now the payload format + a boot probe, not kernels |
+| **Lucebox** | yes, image on host | Qwen 3.8, DS4; **V4.1 on the unmerged `feat/ds41-engine` branch** | The only V4.1 path. **Refuses layer split and layer-major prefill for V4.1** |
+| **radiance-vLLM** (drwolfen) | yes — served frank-primary 2026-09-27 → 09-30 | welded to `qwen4_exp`; PLE / QSA / packed-HC live inside that model class | 12–20 person-days per architecture |
+| **FreeToken-ROCm** | partial | block-FP8 safetensors only (modelopt); known `CpuMoeExecutor` issue under ROCm | cannot load a GGUF |
+| **llama.cpp** | yes | GLM-5.3 confirmed by a community R9700 guide; DS-V4.1 CED unverified | no expert LRU; **8 CPU cores** make CPU expert placement the wrong trade |
+| **TokenSpeed** | claimed | GLM-5.x incl. a 5.3-Flash recipe | NVIDIA-first, unverified on gfx1201 |
+| **KTransformers** | no | GLM-5.3 tutorial exists | built for Intel AMX + NVIDIA |
+| **radiance** (StillDeadcode — the standalone C++/HIP engine, NOT the vLLM fork of the same name) | image `stilldeadcode/radiance` prebuilt for gfx1201 | qwen35 dense/MoE at fp8, qwen4exp quantised-at-convert, llama-shaped | see the probe below: our FP8 payload is REFUSED; tiers, MTP and vision native |
+
+### radiance (standalone) — 2026-10-05 probe
+
+`codeberg.org/StillDeadcode/radiance` — Deadcode's from-scratch engine
+(v1.0.13): same author as the `vllm-radiance` fork we ran in September
+([[frank-medium-radiance-mxfp4]]), an unrelated codebase — C++/HIP, no
+Python, kernels/architectures/quantisers as `.so` plugins, flags written
+for two 32 GB R9700s, native Qwen3.8-Flash-Next with `--placement
+expert_tiered`, MTP and vision.
+
+Probe with the DEPLOYED payload (`rad-convert --plan-only`, which stops
+at declare — no GPU touched, no bytes written):
+
+```
+E no architecture plugin claims 'qwen4exp' at quantisation 'fp8'.
+  'qwen4exp' is served at: (unquantised).
+```
+
+So the 173 G orcarouter FP8 file cannot enter this engine at all: the
+qwen4exp plugin takes bf16/f16/f32/GGUF in and every quantisation happens
+at `rad-convert` time by recipe. Not a lock-in to their published quant —
+recipes are per-weight-class rules (`rtn`/`gptq`/`awq`/`autoround`/
+`paroquant`/`ggml`, grids from fp8 128×128 to i4/w4nl/w5nl codebooks,
+custom `--quant` rules go ahead of the file's), and libr4d lists routed
+experts-as-fp8 among the encodings it reads — so today's quantization is
+reproducible **from a bf16 source**, not from our file. Uncensored-family
+bf16 source candidates: `windowsxp811203/Qwen3.8-Flash-Next-Abliterated-GGUF`
+(ships the bf16 GGUF) and `mradermacher/Qwen3.8-Flash-Next-Uncensored-GGUF`
+(i1 quants only). 622 G free fits source + container. A trial is:
+download → `rad-convert` (CPU-side) → stop+mask frank-primary window to
+serve. Caveats on record: `--tp-wire wht6` is a lossy int8 all-reduce
+wire; the recurrent GDN state is f32 by default (144 MiB/sequence) with
+an opt-in f16 that changes generated text; every pool is budgeted
+explicitly — no profiling pass, no utilisation fraction.
+
+The two-R9700 three-owner split (primary + secondary + stream) is **not
+available on this host**: the retired quadlet recorded that
+`LUCE_DS4_MOE_TP_INPROC` is qualified for gfx1151+gfx1201 pairs, *not*
+gfx1201+gfx1201. hip:0 takes dense weights and the hot expert stack;
+`LUCE_EXPERT_STREAM_DEVICE=1` gives hip:1 the streamed-expert slot cache.
+
+## Measured on Dankserver
+
+<!-- MEASUREMENT -->
+
+## See Also
+
+- [[frank-primary-lucebox-ds4]] — the retired Lucebox frank-primary runbook
+- [[Dual-R9700-Cooperation]] — the PCIe/atomics facts behind the board swap
+- [[Dankserver System Specs]] — the host (RAM corrected 2026-09-28)
+- [[frank-primary-r9v-radiance-qwen38-q8]] — the ACTIVE frank-primary

serverAppConfigs-docs/LLM/dankserver-qwen38-fp8-expert-offload-tuning.md

@@ -0,0 +1,724 @@
+---
+type: research
+title: Qwen3.8-Flash-Next FP8 on tcclaviger — expert-offload decode tuning
+description: VERIFIED 2026-09-29 by measurement. The flag-level decode-tuning campaign on the ad hoc qwen-fp8 container (orcarouter/Qwen3.8-Flash-Next-Uncensored-FP8, tcclaviger RDNA4 vLLM fork, 2x R9700): the >35 tok/s target is MET at 37.4-40.6 tok/s by turning prefetch OFF — telemetry measured its precision at 36%, wasting ~half of all gather traffic — while the original 26.8 tok/s incumbent (prefetch on) survived every other challenger. Residual upside is the fork's gather read path (effective ~6.5-9.5 GB/s vs 22.7 GB/s the link delivers at the same slab size).
+tags:
+  - dankserver
+  - r9700
+  - gfx1201
+  - qwen3.8
+  - fp8
+  - expert-offload
+  - speculative-decoding
+  - tuning
+timestamp: 2026-09-29
+---
+
+# Qwen3.8-Flash-Next FP8 — expert-offload decode tuning
+
+> **Question this page answers.** On the tcclaviger RDNA4 vLLM fork serving
+> `orcarouter/Qwen3.8-Flash-Next-Uncensored-FP8` across both R9700s, what
+> runtime flags actually move decode at a 32k working point?
+>
+> **Answer in one line.** **37.44–40.57 tok/s — the >35 target is met — by
+> deleting `--expert-prefetch 1`.** Expert-cache telemetry measured the
+> prefetcher at **36% precision** (36,720 slabs prefetched, 13,369 served):
+> ~half of all gather traffic was useless copies, and the copies it "saved"
+> would have been installs anyway. The prefetch-on incumbent (26.8 tok/s)
+> survived every other challenger measured: gather geometry, spec depth,
+> TunableOp, scheduler batch knobs, concurrency, ngram-only drafting.
+
+This is the ad hoc `qwen-fp8` container on Dankserver (rootless podman,
+image `docker.io/tcclaviger/vllm:latest`, engine
+`v0.29.0.dev0+g2bdbbc8080`), NOT the `frank-primary` systemd service —
+the R9V production unit is untouched, so no stop-and-mask was required for
+any of this. All numbers are decode tok/s measured from the server's own
+`usage.completion_tokens / (wall - ttft)` — SSE chunk counting is invalid
+under speculative decoding because rejected drafts change the accounting.
+
+## Deployment shape (the tuned incumbent)
+
+Serve script on the host (`/home/dank/serve_qwen_fp8.sh`), env hooks in
+CAPS. **PREFETCH must be unset** (omitting the flag; the fork's default is
+off):
+
+```
+MODEL=/home/dank/models/Qwen3.8-Flash-Next-Uncensored-FP8   (~172.8 GiB, 131 shards)
+IMAGE=docker.io/tcclaviger/vllm@sha256:da4029666668b90a0386c1ed23ea370c79916ad4e1a6ad1dd9db098b9fa069ea
+# image ID 3322dbbbc96d; local alias qwen-fp8-runtime:tuned-20260929;
+# the floating :latest tag moved once already — always launch by digest.
+# repo tag of this state: qwen-fp8-tuned-20260929
+--tensor-parallel-size 2  --max-model-len 32768  --gpu-memory-utilization 0.98
+--kv-cache-memory-bytes 2147483648  --max-num-seqs 4  --enable-prefix-caching
+--speculative-config '{"method":"mtp","num_speculative_tokens":4}'
+# NO --expert-prefetch: measured 36% precision, ~50% wasted gather traffic
+```
+
+Published to the host LAN via the serve script's `PORTMAP` hook
+(`PORTMAP=7773:7770`): health and completions answer on
+`http://192.168.0.235:7773` (verified host-side 2026-09-29; 256k config,
+decode 36.49/34.55/35.98 through the published path). Host port 7773 was
+chosen over 7770 because `frank-primary` (7770) is stopped but ENABLED —
+any future start would collide with a published 7770.
+
+Device notes carried from the bring-up: `--device /dev/dri` must be the
+WHOLE directory (individual renderD nodes enumerate but refuse contexts —
+"No CUDA GPUs are available" while device_count() reports 2); the image
+entrypoint execs `vllm serve` itself.
+
+The memory plan this produces (from the engine's own `expert_plan.py` log):
+
+| region | size |
+|---|---:|
+| expert cache space | 15.05 GiB = 6,574 slots x 2.344 MiB |
+|   VRAM-only resident layers | 3.52 GiB (layers 0-2, 1,536 experts) |
+|   swap region | 11.53 GiB (5,038 slots, 111-112/layer) |
+| host-backed experts | 52.74 GiB per rank, 23,040 experts |
+| KV reserve (one max-len request +8%) | 0.36 GiB of the 2 GiB pool |
+| CUDA graphs | 1.93 GiB (21 graphs; 1.32 GiB at `--max-num-seqs 2`) |
+| runtime overhead | 7.6 GiB (non-torch + buffers + activation) |
+
+## Measured arms (2026-09-29)
+
+Decode tok/s at 166 / 1087 / 4064 prompt tokens, median of 3 runs after a
+64-token warm pass, temperature 0, max_tokens 256:
+
+| Arm | 166 | 1087 | 4064 | Acceptance | Verdict |
+|---|---:|---:|---:|---:|---|
+| **MTP4, stock 8x16, PREFETCH OFF** | **40.57** | **37.44** | **37.57** | 54.0% | **NEW INCUMBENT — target met** |
+| MTP4, stock 8x16, prefetch ON | 26.80 | 25.15 | 26.81 | 60.7% | old incumbent (+51%/+40% worse) |
+| MTP4, prefetch ON (re-check) | 25.80 | 24.38 | 26.63 | 47.9% | reproduces (noise band ±1) |
+| MTP4, prefetch ON, gather 16x32 | 24.30 | 22.77 | 24.11 | 50.3% | −9% — ruled out |
+| MTP4, prefetch ON, gather 32x64 | 26.58 | 24.94 | 26.34 | 49.6% | parity — ruled out |
+| MTP3, prefetch ON | 24.47 | 24.41 | 24.44 | 66.7% | −9%, flat in ctx — ruled out |
+| MTP6, prefetch ON | 19.07 | 17.81 | 13.54 | 17.4% | collapses; crosses UVA threshold — ruled out |
+| MTP4, prefetch ON, MNBT 8192 | 29.32 | 24.12 | 23.65 | — | neutral — ruled out |
+| MTP4, prefetch ON, PREFETCH depth 2 | 25.83 | 25.06 | 25.87 | 56.0% | neutral — ruled out |
+| MTP4, prefetch ON, TunableOp w/ offload | 23.59 | (79.82*) | (79.89*) | (96.5%*) | **rejected — corrupts** |
+| 2-way concurrency, prefetch ON (2 streams) | — | — | 15.06 + 23.11 = **25.49 summed** | — | no gain vs 26.8 single |
+| ngram-only drafter | — | — | — | — | fork bug: PLE inputs unprepared |
+| 256k ctx (MAXLEN=262144, KV 4.5 / 4.03 GiB, prefetch off) | 36.65 / 36.48 | 34.21 / 34.13 | 34.77 / **35.02** | 53.0 / 39.4% | boots and serves; band 34.1-36.5 straddles 35 within noise; arena identical (5,648) at both KV sizes |
+
+\* The TunableOp numbers are invalid by construction: the tuned GEMMs
+changed generated behavior (repetition-loop, the server's
+"repetition detected" halt firing at ~140-146 tokens instead of the full
+256, acceptance jumping 60.7% → 96.5%), the 79.8 tok/s figures were
+measured on those halted degenerate completions, and the text-only check on
+a fresh prompt came back 20.01 tok/s — *below* baseline. GEMM autotuning
+that changes output text is broken tuning.
+
+The decode curve without spec decode is dead flat 12.11 tok/s from 304 to
+16,095 tokens — decode is context-independent on this shape, bound per-step
+by the expert gather, not by KV depth.
+
+## Telemetry — the measurement that found the win
+
+`R4D_LRU_TELEMETRY=1` arms per-layer counters; a torch-profiler window
+(`/start_profile` → traffic → `/stop_profile`) makes the fork log and write
+`expert_cache_rank<N>.json` beside the traces. **`torch_profiler_with_stack=false`
+is mandatory in the profiler config** — the default (true) grew a 6.5-min
+window's trace buffer ~40 GB and the kernel OOM-killed the engine core
+(anon-rss 95.2 GB, `oom-kill` in `journalctl -k`). The short window at
+166 ctx (~15 s of traffic) completed cleanly with `with_stack=false`.
+
+The fork's own numbers (102 steps, both ranks identical):
+
+```
+hit rate 0.8245 (137,084/166,262)   host installs 15,814
+read-through 13,364                 prefetched 36,720
+served by prefetch (migrations) 13,369
+```
+
+- **The LRU is good**: 82.45% hit rate overall; layers 0-2 pinned at
+  100%; swap-region layers 78-81%. The "arena holds only 27% of experts
+  → 73% miss" arithmetic was wrong — temporal reuse carries it.
+- **The prefetcher is the poison**: 36,720 slabs moved to serve 13,369
+  (36% precision). The 23,351 unused copies are ~0.7 GB/step/rank of pure
+  waste — roughly half of all gather traffic — and the 13,369 "savings"
+  would have been installs anyway. Depth-1 prefetch also cannot cover the
+  ~2,400-slab per-step working set with its 16 slots/layer (768 total).
+- Per-layer-step at a swap layer: ~30.6 routed, 24.8 hits, 3.3 installs,
+  2.6 read-through, 8.0 prefetched of which 1.7-3.3 useful.
+
+## The wall, measured
+
+- **PCIe link (pcie-link-probe):** GPUs `03:00.0`/`07:00.0` LnkCap/LnkSta
+  32GT/s x16; root ports `00:01.1`/`00:01.3` cap x8. Effective **Gen5 x8 =
+  31.5 GB/s/direction**. `rocm-smi pcie_link_speed: 25` is the ASPM idle
+  downshift, not the loaded rate.
+- **H2D (HIP probe, pinned):** 2.345 MiB (the engine's exact expert slab)
+  **22.74 GB/s**; 16 MiB 28.19; 1 GiB 28.74 (91% of ceiling). Pinned ≈
+  pageable — staging is not the lever. Single-transfer latency at slab size
+  105.7 µs vs 78 µs ideal → ~21% per-transfer overhead.
+- **Engine's effective gather rate:** ~6.5-9.5 GB/s = **30-42% of what the
+  link delivers at the same size**. The bottleneck is the fork's read path
+  (chunk/lane geometry, per-transfer cost), not the wire.
+- **Duty cycle:** peak GPU util 100%, smooth ~29% with prefetch on — each
+  step is ~100 ms of compute then hundreds of ms waiting on expert
+  streaming. rocm-smi reports *utilization*, which reads ~0 while weights
+  load during an arm's reload; VRAM is the honest panel there.
+
+## What the remaining flags cannot do (all measured)
+
+- **Gather geometry** — 16x32 and 32x64 measured no better despite the
+  fork's "tuned at the Gen4 ceiling" comment; the constants are already at
+  parity on Gen5 x8. The read-path gap is structural, not a tuning knob.
+- **Spec depth** — 4 is the optimum both ways (with prefetch on). Depth 6
+  additionally proves the UVA read-through path (`step tokens >
+  R4D_LRU_UVA_TOKENS=5`) is *slower* than copy-in.
+- **Prefetch anything** — depth 2 neutral, depth 1 the new negative; the
+  tier is 3x too small for the working set and the predictor is 36%
+  precise. `expert_prefetch_k/budget/slots` exist (defaults 20/14/16) but
+  the precision result makes tuning them pointless.
+- **Arena growth via shrink** — `UTIL=0.99` fails boot (`Free memory on
+  device cuda:1 (31.29/31.86)`); `--max-num-seqs 2` fails CUDA-graph
+  capture (`Constraints violated (L['query_start_loc'].size()[0])`); and
+  the MNBT 8192 arm already proved a ~1 GiB arena shift is decode-neutral
+  in both directions.
+- **Concurrency** — under TP=2 both streams' tokens are computed on both
+  cards, so each rank gathers the union of both streams' misses; the links
+  do not add capacity and the shared arena thrashes. Summed 2-way (25.49)
+  measured *below* single-stream (26.8) with prefetch on.
+- **ngram-only drafting** — fails at engine init:
+  `RuntimeError: PLE inputs were not prepared`; the fork prepares this
+  model's PLE inputs only on the MTP spec path. Needs a fork patch, not a
+  config.
+- **V-only q8 KV** — stock `--kv-cache-dtype` offers fp8 variants only
+  (K+V together); a V-only q8 needs a fork patch.
+
+## Fork bugs / traps found (for the engine authors or a future patch)
+
+1. `CLAV_TUNABLEOP_SWEEP=1` is silently skipped whenever expert offload is
+   active unless `CLAV_TUNABLEOP_WITH_OFFLOAD=1` is also set
+   (`tunableop_warmup.py`).
+2. TunableOp-with-offload, once forced on: +6.7 min warmup, no persistent
+   cache (container-local), and corrupt numerics (repetition loops,
+   early-EOS halts, invalid measurements). Rejected outright.
+3. The torch-profiler stop path hangs the workers on ROCm when
+   `torch_profiler_with_stack` is true (default): >10 min hangs on two
+   attempts while the AsyncLLM wrapper reports "stopped successfully" in
+   1 s; the default stack-tracing grew a 6.5-min window's buffer ~40 GB
+   and OOM-killed the engine core. With `with_stack=false` the stop
+   completes and the telemetry report runs. The telemetry counters remain
+   reachable ONLY through that profiler stop — no direct RPC trigger
+   exists (`telemetry_report` has exactly one caller, `gpu_worker.py`).
+4. ngram spec method does not prepare PLE inputs (above).
+
+## Where the headroom lives now
+
+The 35 tok/s target is met (37.4-40.6) with prefetch off. What is left:
+
+1. **Fork patch to the gather read path** — the ~6.5-9.5 → 22.7 GB/s
+   software gap at slab size. If the gather reached wire rate, decode
+   lands in the 60+ tok/s band before acceptance effects. Code work in
+   `expert_lru.py`'s gather (`R4D_LRU_GATHER_*` is read at import; the
+   copy-in path itself is compiled — see `/app/r4dhip/r4d.so`).
+2. **256k ctx — measured, final**: boots at `MAXLEN=262144` with
+   `KV_BYTES=4831838208` or `4328521728` alike — a 262,144-token window
+   needs **4.0 GiB** KV (hybrid attention + mamba pages ≈ 16 KiB/token,
+   not the 32k plan's 9,190 B/token; 3 GiB boot-refuses and the engine
+   estimates max len 184,224 at that pool; the exact-4.0 override loses
+   ~20 MB to accounting and boots as 3.98 < 4.0). Decode band
+   **34.1-36.5** across both KV sizes — the arena is **identical (5,648
+   slots)** at 4.5 and 4.03 GiB (the planner buckets; the trim went to
+   margin, not slots), so the window costs ~10% decode vs the 32k config
+   and no KV trim recovers it. The alternative that keeps the full
+   37.4-40.6 band is `MAXLEN=184224` at 3 GiB KV. Correction: the planner
+   sizes the arena AFTER the KV override (the earlier "before" reading was
+   wrong) — the arena is what remains.
+3. **2+ slots at long ctx** — the KV pool post-spec holds ~66k tokens at
+   2 GiB (~2x 32k streams); concurrency summed *below* parity at 4064 ctx,
+   so slots are a capacity feature, not a throughput one. Two full 256k
+   streams need ~8-10 GiB of KV — that fits the arithmetic only by cutting
+   the arena below usefulness; slots-at-256k is a fork-patch or
+   smaller-payload question.
+
+## Reproducing
+
+Host-side scratch scripts (2026-09-29): `/home/dank/serve_qwen_fp8.sh` (the
+serve script with all hooks), `/home/dank/mtp_measure.py` (usage-counter
+decode measure, 3 depths), `/home/dank/mtp_measure_short.py` (single depth,
+for profile windows), `/home/dank/bench2way.py` (concurrency), plus the arm
+scripts `arm_*.sh` / `sweep_*.sh` and the detached runner `/home/dank/job.sh`
+(outputs under `/home/dank/jobs/NAME.out|.rc|.done`). The measure scripts
+are `podman cp`'d into the container after every recreate (the serve script
+re-copies them; a recreate wipes container-local files).
+
+## 2026-10-01 — production cutovers (qwen FP8+MTP4, then bf16 no-spec)
+
+Two same-day cutovers moved this page's findings into production:
+
+1. **Qwen FP8 + MTP4 replaced the failed GLM/DFlash2 experiment** (the GLM
+   window is documented in [[frank-primary-glm53-flash-w4a16]]; git tag
+   `frank-primary-fp8-mtp-20261001` is the last commit of that shape). The
+   production trio took MAXLEN 200000 x 3, fp8 KV, MTP4, prefetch off —
+   measured 37.4-40.6 tok/s at ~54% acceptance.
+2. **The overnight redo campaign re-measured EVERY lever at matched
+   geometry** (prefetch OFF everywhere, MAXLEN 200000/SEQS 3/fp8 baseline)
+   and concluded the day: the incumbent cutover to **bf16 KV, no
+   speculative decode** (Frank's call).
+
+### What the redo campaign measured (all at shallow depth, 166/1087/4064)
+
+| lever | verdict |
+|---|---|
+| prefetch (re-confirmed) | the whole night's +40-51% lives in deleting it; the old 12.11 "no-spec baseline" was itself prefetch-taxed |
+| gather geometry | **R4D_LRU_GATHER_CHUNKS=32 / LANES=64 = +18.7%** (49.25 @4064), the win GROWS with depth (39.5@166 -> 54.1@1087); g8x16 mild, g64x128 past peak |
+| KV dtype | bf16 (`auto`) +4% decode on fp8-era numbers but halves attention-KV capacity (measured 14,047 vs 7,926 B/token; mamba states unaffected) |
+| spec family | no-spec fastest once the tax was gone (41.46/41.29/41.47); MTP4's drafter reserves ~2.6 GiB/rank beside the pool; untaxed MTP re-measures never completed |
+| MNBT | 512 doubles TTFT (2.46 s vs 1.36 s @4064); 1024-4096 interchangeable; 12288 -2.5% + arena; **locked 1024** |
+| UVA | 2048 POISON (-39%); 8192 ~parity |
+| arena | flat ±1.4% from ~6,830 down to ~5,500 slots; knee below |
+| six slots | every SEQS-6 KV sizing refused at engine init |
+
+Depth caveat: every number above is at a near-empty KV. Deep-context
+behavior (the gather win growing with depth, bf16's decode cost at depth)
+is only sampled at 3 depths; a 64k-192k-depth probe is the flagged gap.
+
+### The deployed shape (2026-10-03 shape G: 3 slots x 200704, fp8, precise memory, MNBT 2048, arena at the flag ceiling)
+
+```
+FRANK_PRIMARY_MAXLEN=200704   200704-token window
+FRANK_PRIMARY_SEQS=3          --max-num-seqs 3 (SEQS=2 CANNOT boot - below)
+FRANK_PRIMARY_KV_BYTES=3166500000   PER RANK (fp8, 7,892 B/token -> 400,700 tokens)
+FRANK_PRIMARY_KV_DTYPE=fp8    --kv-cache-dtype fp8 (quantized)
+FRANK_PRIMARY_MNBT=2048       --max-num-batched-tokens 2048 (arena-free, see shape F)
+FRANK_PRIMARY_UTIL=0.982      budget = total x util, the law (1.0 unbootable, see below)
+FRANK_PRIMARY_CUDAGRAPH_BASE_MB=0 / _PER_GRAPH_MB=12  -e VLLM_CUDAGRAPH_MEM_*
+FRANK_PRIMARY_PRECISE_MEMORY=1      --expert-precise-memory (overhead 7.60 -> 5.25 GiB)
+FRANK_PRIMARY_EXPERT_OFFLOAD_MEM=120  --expert-offload-mem (host-region cap, explicit)
+FRANK_PRIMARY_GPU_MAX_HW_QUEUES=2   -e GPU_MAX_HW_QUEUES=2 (HSA queues/GPU)
+FRANK_PRIMARY_CACHE_DIR=/home/dank/.cache/frank-primary   mounted at /root/.cache
+R4D_LRU_GATHER_CHUNKS=32 / R4D_LRU_GATHER_LANES=64  (container env)
+NO --speculative-config, NO --expert-prefetch
+FRANK_PRIMARY_NO_PREFIX_CACHING=1  --no-enable-prefix-caching (2026-10-05, the QFN align-mode GDN state fault — below)
+```
+
+### 2026-10-05 — prefix caching OFF: the fork's QFN align-mode GDN state fault
+
+**Symptom (Frank, 2026-10-04):** output turned repetitive/corrupted at
+~40k tokens into a long conversation; short sessions stayed clean.
+
+**Evidence (live `podman logs frank-primary-qwen`, the 2026-10-04 18:19
+boot):** one 0.6 s burst at 18:46:07 PDT — 5.6 h into service, one
+running request, and the engine's own 18:46:08 log line reading
+"Prefix cache hit rate: 48.1%". 48 `gdn_hip` faults, then the
+process-lifetime silence cap:
+
+```
+[gdn_hip] ERROR gdn_decode_r: state out of ssm-cache range: max|S|=0
+  with non-finite values (slot=42 head=3 limit=3.40282e+38) fault #1-23
+[gdn_hip] ERROR gdn_decode_r: ... (slot=42 head=8 ...) fault #24-46
+[gdn_hip] ERROR: state fault report limit (32) reached; further faults
+  are counted silently
+```
+
+One physical SSM slot (42), two heads (3 and 8 — one per TP rank), 23
+GDN layers each, the state non-finite (NaN). The guard's own text in
+this build: *"known Qwen3.8-Flash-Next (QFN) flaw under prefix caching
+and is being actively worked; output quality for this request may
+degrade. Mitigation: --no-enable-prefix-caching."*
+
+**Mechanism.** With prefix caching on, the fork defaults QFN to
+`mamba_cache_mode=align` (boot log: "Mamba cache mode is set to 'align'
+for Qwen4ExpForConditionalGeneration … when prefix caching is enabled";
+`config/cache.py`: "none" is the no-prefix-cache mode, "all" is
+deprecated — align is the only sound mode WITH the flag). Align mode
+checkpoints each sequence's GDN (Gated DeltaNet) recurrent state into a
+mamba page per block and RE-SEEDS the state from that page when a
+request resumes from a cached prefix. The block size is raised to
+**3136 tokens** for the page-size equality (boot log: "Setting
+attention block size to 3136 tokens to ensure that attention page size
+is >= mamba page size" + "Padding mamba page size by 0.13% … exactly
+equal"). The corrupt state landed at the **13th boundary: 13 × 3136 =
+40,768 tokens** — the reported "~40k" mark. The NaN persisted in the
+mamba page, so every later turn of that conversation re-seeded the same
+non-finite state for those two heads (silently — the fault-report cap
+was already spent) while fresh short sessions looked fine.
+
+**RAM ruled out the same day** (the 2026-10-04 17:17 host reboot is when
+the current memory speed took effect): zero MCE/EDAC kernel records
+since boot; an 8 GiB in-band pattern check stable over 60 s under live
+load; 4800 MT/s is the JEDEC DDR5 floor on 5600 MT/s-rated DIMMs (not an
+overclock); and a two-head, single-slot NaN state is a state-path fault,
+not scattered finite bit corruption.
+
+**Fix.** `FRANK_PRIMARY_NO_PREFIX_CACHING=1` → `--no-enable-prefix-caching`
+(the fork-named mitigation; the flag renders in both directions from the
+knob so a fork default change cannot silently flip the contract). Cost:
+no prefix reuse — every turn re-prefills the whole history instead of at
+most 3135 tail tokens. A silently corrupted conversation is worse than a
+slower first token.
+
+**Verified 2026-10-05 (deploy b820d34):**
+
+- Engine argv carries `--no-enable-prefix-caching`; health 200; warm boot
+  `init engine … took 76.57 s` (the mounted compile cache is intact).
+- Boot plan unchanged, as predicted: expert cache **17.77 GiB / 7,762
+  slots** (the flag touches no plan term), CUDA-graph estimate 0.07 GiB
+  vs 0.08 GiB captured (inside the 12 MiB/graph charge).
+- `GPU KV cache size: 424,678 tokens` — **up** from 401,408 at the same
+  3,166,500,000 B ask: the pool now prices ~7,456 B/token (was 7,891.8),
+  the mamba/side per-token term dropping once the per-block checkpoint
+  pages are gone; the 3136 attention block-size raise nonetheless stays
+  (the page-size invariant is enforced in mode "none" too). Frank's
+  400,700-token floor is cleared with ~24k of headroom.
+- **Zero `gdn_hip` lines** in the boot-and-traffic window (only the two
+  INFO module banners mention the name).
+- Boundary probe (single request, no prefix reuse): prompt 39,280 +
+  completion 1,491 = **40,771 total tokens — across the 40,768 line**;
+  the task (write 1–400, one per line) completed coherently: 399 integers
+  parsed, 0.997 monotone, tail `… 398 399 400`. Pre-fix, this boundary is
+  exactly where the conversation degraded (the 13th 3136-page re-seed).
+- e2e: 32/33 pass; the one failure is the admin spec's
+  192.168.0.112:8989 leg with FrankBook-Pro asleep (environmental,
+  unrelated to this change; re-run `tests/e2e/run.sh` once the mini is up).
+
+**RESULT — REVERTED THE SAME DAY.** With every mechanical check green
+(zero faults, coherent 40,771-token probe), live long conversations
+STILL degraded at ~40k on the running endpoint (Frank, direct
+observation). The container serving that window (`--no-enable-prefix-
+caching`, booted 2026-10-05 12:47 PDT) logged ZERO `gdn_hip` faults
+while the corruption reproduced — so this recurrence is **fault-free**:
+the NaN-state guard did not fire, which the re-seed theory requires.
+Open hypotheses, in order: (1) the poisoned transcript is being re-fed —
+corrupted assistant turns already in the history keep producing
+corrupted continuations even with perfect state, so the same
+conversation cannot distinguish fixed from unfixed; (2) a second
+long-context mechanism with no guard coverage — fp8 KV degradation at
+depth is the next suspect (a bf16-KV boot costs ~2× the pool per token
+and would settle it). Prefix caching is **back ON**
+(`FRANK_PRIMARY_NO_PREFIX_CACHING` commented out — the cost bought no
+fix). The knob and the serve branch stay for a future fork build with a
+fixed GDN path. Next diagnostic: reproduce on a FRESH conversation; if
+it still breaks at ~40k with zero faults, A/B the KV dtype.
+
+**RESOLVED (2026-10-05 evening): it was the RAM.** Frank dropped the
+memory from 4800 to **4600 MT/s** (in effect since the 2026-10-05 12:21
+PDT host boot) and the corruption stopped: zero `gdn_hip` ERROR lines in
+the container since, zero MCE records, and the live agent session on
+this endpoint past 140k context, "comfy cozy" (Frank). Marginal DDR5 on
+four DIMMs is a common cause for BOTH eras: the 2026-10-04 NaN burst
+(corrupted weight/activation traffic poisoning a GDN state write — the
+guard-visible path) and the fault-free ~40k degradation (silent flips in
+KV/expert data, which no guard covers on non-ECC memory). The re-seed
+theory is therefore UNPROVEN, not disproven — the entire caching
+experiment ran on marginal RAM and was confounded on both arms.
+Post-mortem on the rule-out: consumer non-ECC RAM raises no MCE for a
+silent flip; a 60 s / 8 GiB pattern sample covers ~0.5 % of 246 GiB;
+"4800 is the JEDEC floor on 5600-rated DIMMs" is a rating, not a
+stability proof at four-DIMM loading. The revert test was the decisive
+instrument and it pointed at the RAM all along. Prefix caching stays ON
+(zero faults under heavy re-send traffic since the speed drop); the knob
+stays for a future fork fix. The fp8-KV A/B is NOT needed.
+
+### Shape F/G (2026-10-03): where the free VRAM actually is, and what forcing it cost
+
+`rocm-smi` showed **3.57 GiB free per R9700** at the deployed shape E (28.28 GiB
+used of 31.86 GiB, after hours of real traffic — so it is the high-water mark,
+not an idle artifact). Growing the expert arena into it is NOT a matter of
+raising `--gpu-memory-utilization`, because the arena is
+
+```
+arena = budget - non-expert - KV reserve - CUDA graphs - runtime overhead
+budget = device_total x gpu_memory_utilization          # expert_plan.py:567
+```
+
+and every term on the right is either fixed or a lie:
+
+| term | plan charges | measured reality | reachable from our side? |
+|---|---|---|---|
+| `budget` | 31.22 GiB (0.98) | field is `le=1` so 31.86 GiB is the field max, but 1.0 is UNBOOTABLE (below) | **yes** — shipped 0.982 |
+| `non-expert weights` | 6.28 GiB | checkpoint census | no |
+| `KV reserve` | 1.91 GiB (registry 9,190 B/token x 200704 +8%) | real pool bills 2.93 GiB at 7,892 B/token | no (registry constant) |
+| `CUDA graphs` | 0.55 GiB = 256 MiB + 52 MiB x 6 | `Graph capturing finished in 3 secs, took 0.31 GiB` | **yes** — `VLLM_CUDAGRAPH_MEM_BASE_MB` / `_PER_GRAPH_MB`; `VLLM_CUDAGRAPH_MEM_ESTIMATE_MB` pins the whole term outright (checked first, pair ignored when set) |
+| `runtime overhead` | 5.25 GiB (`--expert-precise-memory`) | far lower at idle | no (registry constant) |
+
+Shape F took the graph term (0.55 -> 0.32 GiB) and nudged util to 0.982:
+**arena 17.52 GiB / 7,653 slots measured**, +131 slots. Shape G takes the
+graph term further (0.32 -> 0.07 GiB) at the same util.
+
+**Shape G takes the graph term to the measured floor, util to the law.** The
+graph base pool is the actual over-charge: shape F's own boot reported
+`Graph capturing finished in 4 secs, took -0.06 GiB` — a **negative** capture
+delta, so the 256 MiB base was fiction. Charging `0 base + 12 MiB/graph`
+(0.07 GiB for the shipped 6-graph plan, and still per-graph so an 18-graph
+spec-decode shape pays 0.21 GiB) lands the plan at budget 31.29 / graphs 0.07 /
+**arena 17.77 GiB / 7,762 slots measured**. The KV ask went to
+`3,166,500,000` B/rank, which the engine's own 7,891.8 B/token price lands at
+**401,408 tokens measured / 2.00x concurrency** — the first shape holding two
+full-length conversations with real spare. (The ask/landed delta is block
+alignment: 196 x 2048. The price is shape-dependent, 7,701 B/token at 262144.)
+
+**UTIL 1.00 is not a knife-edge on this fork, it is UNBOOTABLE — measured
+2026-10-03.** One deploy took the field maximum to buy the last 0.57 GiB of
+arena; both workers refused init BEFORE loading a single weight, twice:
+`ValueError: Free memory on device cuda:N (31.29/31.86 GiB ...)` — the worker
+demands free > budget and budget == total, so `le=1` on the CLI field is a
+trap: the engine dies in a 2-restart crash loop with zero payload progress.
+So `0.982` is a LAW, not a preference, and the "why 0.982" answer now has two
+proofs: this ValueError, and the historical ceiling deaths (the 2x262144 bf16
+boot at 0.13 GiB free; the 5x262144 boot at "0 bytes free" on a real
+995-token request). Everything beyond ~7,912 slots needs the registry edited
+inside the image (the fork's log line says exactly that: *"Update the
+registry entry if this drifts"*), not a deploy.
+
+**The host region was the last arena-plan term still autofitting per boot —
+now it is a literal (2026-10-04).** With no `--expert-offload-mem` the planner
+binds the host region to `MemAvailable` AT BOOT minus `PLE 47.68 - 2 x 7.0
+process - 5.0 api/engine - 5% margin`: the 2026-10-05 boot read 237.97 GiB and
+offered 159.39 GiB usable, while the same host reads **75.4 GiB available while
+running** — the bound tracks whatever the co-tenants hold at that moment. The
+demand is 105.48 GiB (52.74 GiB/rank x 2, 23,040 swap-target experts), and a
+boot where the bound drops near it silently converts next-hottest experts to
+VRAM-only (the `pinned_extra` warning is the only trace).
+`FRANK_PRIMARY_EXPERT_OFFLOAD_MEM=120` pins it: measured demand + 14% headroom,
+on a host with 246.1 GiB total. Semantics verified in the shipped image
+(`offloader/expert_plan.py`): the cap binds as the bound, an exceedance only
+warns, and the plan rewrites the field to the cap it met (+1 MiB/rank).
+
+**MNBT 1024 -> 2048 is free on every axis the plan can see.** The registry
+overhead model is `base + 0.22 GiB x max(0, (nbt - 4096)/1024)` anchored at
+nbt 4096, so any rung up to 4096 charges the identical 5.25 GiB; the fitted 5.25
+was itself fitted at nbt 4096, which means 2048 is covered by an activation
+figure measured at a **larger** batch. SEQS 3 boots at both 1024 and 2048
+(measured 2026-10-02 — the `query_start_loc` `ConstraintViolationError` is a
+SEQS property). The costs are a one-time recompile (the compile **range** is
+part of the cache key, so `FRANK_PRIMARY_CACHE_DIR` pays once) and a wider
+prefill chunk, which the 09-29 sweep found decode-interchangeable across
+1024-4096 while 512 doubled TTFT at 4064 ctx (2.46 s vs 1.36 s).
+
+**`--expert-precise-memory` is the arena lever that does not touch the window.**
+It attacks the planner's overhead term (shape F above attacks the graph term —
+the third and last term this repo can reach), and the flag swaps the default
+registry figure for the fitted one. Measured
+2026-10-02 by a boot probe on the shape-D base (production stopped, candidate
+booted AS the service; numbers are the engine's own plan lines):
+
+| plan line | shape D | **shape E** |
+|---|---|---|
+| `runtime overhead (non-torch + buffers + activation)` | 7.60 GiB | **5.25 GiB** (`precise (gfx1201 TP2 nbt4096/seqs16/len262144)`) |
+| `expert cache space` | 14.87 GiB | **17.22 GiB** |
+| slots | 6,496 | **7,522** (+1,026, +16%) |
+| `swap region` | - | 13.7 GiB / 5,986 slots, 133-134 per layer |
+| boot | - | healthy in 420 s, no OOM |
+
+`'expert_precise_memory': True` also appears in `non-default args`, so a silent
+argparse drop is visible in the boot log. The flag lands **in the arena, not the
+pool** - `--kv-cache-memory-bytes` is untouched at 3.15e9/rank (option A), because
+the pool is not an arena lever and cutting it further would only cost
+concurrency: a 2.2e9 pool was probed in the same window and reports 277,686
+tokens / **1.38x**, so it is the conservative fallback if free VRAM ever binds,
+not the shipped value.
+
+**The compile cache is a boot-cost fix, and it was worth 173 s per boot.**
+Measured 2026-10-02: the image ships **no `/root/.cache` at all** (a fresh
+container: `No such file or directory`), and after a boot that directory holds
+`vllm` at **1,154,714,333 B in 4,665 files** plus `comgr` at 229 MB, all inside
+the container's ephemeral layer. Every
+`podman rm -f` - i.e. every deploy, restart and reboot - discards it, and the
+engine re-pays:
+
+```
+Using cache directory: /root/.cache/vllm/torch_compile_cache/b265bd396a/rank_0_0/backbone
+Dynamo bytecode transform time: 4.99 s
+Compiling a graph for compile range (1, 1024) takes 173.19 s
+saved AOT compiled function to /root/.cache/vllm/torch_compile_cache/torch_aot_compile/…
+torch.compile took 181.20 s in total
+init engine (profile, create kv cache, warmup model) took 217.26 s (compilation: 181.20 s)
+```
+
+Mounting a host directory at `/root/.cache` makes every later boot load the AOT
+artifact instead. The mount is deliberately the **parent**, not
+`/root/.cache/vllm`: `VLLM_CACHE_ROOT` defaults to the `vllm` child, and
+`comgr` is the HIP-kernel JIT cache, so one mount persists both — and it hides
+nothing, because the image ships no `/root/.cache`. The cache key
+includes the compile range, so the MNBT choice costs exactly one compilation.
+Entries are version- and config-hashed: a stale entry is a miss, never a wrong
+answer.
+
+**Proven live on the deployed endpoint, 2026-10-02, consecutive boots of this
+exact shape:**
+
+| boot | engine log |
+|---|---|
+| cold (deploy) | `Compiling a graph for compile range (1, 1024) takes 179.24 s` → `torch.compile took 188.26 s in total` |
+| warm (restart) | `Directly load the compiled graph(s) for compile range (1, 1024) from the cache, took 2.143 s` → `torch.compile took 3.97 s in total` → `init engine (profile, create kv cache, warmup model) took 41.13 s (compilation: 3.98 s)` |
+
+The host directory held **1.4 GB in 4,707 files** after the cold boot. Time to
+`/health` is dominated by the 172.8 GiB payload load (the warm boot reached
+health in 490 s), so the win is the ~184 s of inductor/AOT build rather than a
+proportional cut in boot time.
+
+`GPU_MAX_HW_QUEUES=2` caps the HSA hardware queues per GPU. Verified a real knob
+rather than a no-op - the literal is read out of
+`/opt/rocm/core-10.0/lib/libhsa-runtime64.so` - and shipped deliberately
+alongside the 32x64 gather geometry. It is **not** A/B measured.
+
+**The pool, not the slot count, sets how many conversations live at once.**
+Measured live at this shape: `GPU KV cache size: 398,658 tokens, Maximum
+concurrency for 200,704 tokens per request: 1.99x` (3,146,129,408 B reserved).
+The per-token price is **shape-dependent**: the 262144 shape priced at 7,701
+B/token (6,100,000,000 B -> 792,069 tokens), this pool at 7,892 - ~2.5% dearer,
+so a concurrency figure never extrapolates across geometries. The endpoint
+advertises 3 slots: a third LONG request queues rather than being refused, and
+shorter requests use the spare blocks. Verified: three concurrent 67-token
+requests answered in parallel, ~4.1 s each. Two contexts at 200704 are a
+capacity contract, not a throughput one (2-way concurrency summed BELOW parity
+at 4064 ctx).
+
+#### `--max-num-seqs 2` is impossible on this model (measured 2026-10-02)
+
+Four single-variable boot probes against the then-deployed shape, production
+stopped for each:
+
+| MAXLEN | SEQS | MNBT | result |
+|---|---|---|---|
+| 262144 | 2 | 1024 | **Engine core initialization failed** |
+| 262144 | 2 | 2048 | **Engine core initialization failed** |
+| 262144 | 3 | 1024 | healthy - 792,069 tokens, 3.02x, arena 6,248 |
+| 262144 | 3 | 2048 | healthy - the then-deployed incumbent |
+
+The failure is `torch.fx.experimental.symbolic_shapes.ConstraintViolationError:
+Constraints violated (L['query_start_loc'].size()[0]) ... specialized it to be
+a constant (3)`, raised from AOT `build_guards(strict_error=True)` during
+`determine_available_memory -> profile_run -> _dummy_run`. `model_runner.py`
+sizes that input buffer `max_num_reqs + 1` and `_dummy_run` sets
+`num_reqs = min(max_num_batched_tokens, max_num_reqs)`, so the width is
+`SEQS + 1` for ANY `MNBT >= SEQS`: 3 at SEQS 2, 4 at SEQS 3. **MNBT is not a
+lever here** - it only moves the compile RANGE ((1,1024) vs (1,2048)), which
+is why the identical error appears at both rungs. `--enforce-eager` bypasses
+the path at the cost of CUDA graphs. The same failure class killed the
+2026-10-01 bf16 2 x 262144 shape.
+
+#### The arena lever is `max_model_len`, not `--kv-cache-memory-bytes`
+
+The fork's planner sizes expert cache as `budget - non-expert - KV reserve -
+graphs - overhead`, and the KV reserve is registry-priced at **9190 B/token x
+max_model_len** - not from the `--kv-cache-memory-bytes` value:
+
+| MAXLEN | KV reserve | arena |
+|---|---|---|
+| 262144 | 2.48 GiB | 14.30 GiB / **6,248 slots** |
+| 200704 | 1.91 GiB | 14.87 GiB / **6,496 slots** |
+
+D(reserve) 0.57 GiB = D(arena) 0.57 GiB = +248 slots. The 6.1e9 -> 3.15e9
+pool cut (2.75 GiB) bought **zero** arena slots. It buys runtime HEADROOM,
+which is still worth having - the shape it replaced died with `0 bytes free`
+on a real 995-token request. Grow the arena by lowering `max_model_len`, or by
+taking the overhead term with `--expert-precise-memory` (below); the pool is the
+headroom lever only.
+
+#### The other arena lever: `--expert-precise-memory` (overhead 7.60 -> 5.25 GiB)
+
+The flag is now ON. It was refused on 2026-10-01 for a reason that does not
+generalize: the 2 x 262144 bf16 shape was already at the ceiling (weights 6.28 +
+pool 6.80 + graphs 0.55 + arena 16.65 + non-torch ~1.45 = 31.73 of 31.86 GiB,
+NRestarts=3 at ~6.7 min each), so the 2.35 GiB the flag moves into the arena was
+the last stack over the line, not the cause. The plain bf16 2 x 262144 boot also
+did not stay up: 0 bytes free at runtime, and a real client request (995-token
+prompt, `max_tokens=261149`) killed the worker with `Tried to allocate 50.00 MiB
+... 0 bytes free`.
+
+At 200704 x 3 the flag fits with room and is worth **+1,026 slots (+16%)**. What
+made it affordable is the window trim, not a pool cut: the arena is
+DEVICE-resident, the pool is not part of the budget's arithmetic, and the 200704
+window's KV reserve (1.91 GiB) plus a smaller attention footprint is what the
+plan needed. The pool stays at 3.15e9/rank deliberately (option A) - cutting it
+further would cost concurrency (2.2e9 -> 1.38x) and buy no slots.
+
+Boot-probed 2026-10-02 (transient systemd units at the production uid +
+memlock, full-log capture; the rootless bench probe had masked the production
+memory environment). The night searched two concurrency contracts:
+
+| rung | result |
+|---|---|
+| 3 x 262144, bf16, KV 11,047,010,304 | **OOM at graph capture** — `Tried to allocate 76.00 MiB ... 0 bytes free`, 30.43 GiB of 31.86 GiB already allocated by PyTorch |
+| 3 x 245760, bf16 | **OOM identically** |
+| 3 x 237568, bf16 | PASS — KV 718,606 tokens |
+| 2 x 262144, bf16 + `--expert-precise-memory` | **DEPLOYED then crashed** — 0.13 GiB free at boot, NRestarts=3 (the ceiling, not the flag) |
+| 2 x 262144, bf16 (no precise flag) | did not stay up either: 0 bytes free at runtime, worker killed by a real request |
+| 3 x 200704, fp8, pool sized for 2 contexts | **DEPLOYED 2026-10-02 (shape D)**: 398,658 tokens = 1.99x concurrency, arena 14.87 GiB / 6,496 slots, the 2.75 GiB/rank pool cut returned as runtime headroom |
+| **3 x 200704, fp8, + `--expert-precise-memory`** | shape E (2026-10-02): same 1.99x pool, overhead 5.25 GiB, arena 17.22 GiB / 7,522 slots, boot healthy in 420 s |
+| 3 x 200704, fp8, MNBT 2048, graph term 0.32 GiB, util 0.982 | shape F (2026-10-03): arena 17.52 GiB / **7,653 slots** measured, budget 31.29, `Estimated CUDA graph memory: 0.32 GiB`, KV unchanged 398,658 |
+| 3 x 200704, fp8, graph term 0.07 GiB, util **1.00**, KV 3.1665e9 | **DID NOT BOOT (2026-10-03)**: `ValueError: Free memory on device cuda:N (31.29/31.86 GiB)` on both ranks, twice, zero weights loaded — the worker demands free > budget and budget == total. UTIL 1.0 is unbootable, full stop |
+| **3 x 200704, fp8, graph term 0.07 GiB, util 0.982, KV forced to 400,700** | **DEPLOYED 2026-10-03 (shape G)**: arena **17.77 GiB / 7,762 slots** measured (+109 over F, +240 over E), KV **401,408 tokens / 2.00x** concurrency measured, capture 0.31/0.10 GiB vs the 0.07 charge (under-charged, boots and holds), driver used 30.97 GiB -> **0.89 GiB real free per card** held through a live 4,096-token generation (~38-39 tok/s single-request, +13% vs the 33.99 shape-E baseline); rollback is UTIL or KV_BYTES down a notch |
+
+Six full-length contexts cannot fit at any KV dtype: the 6-seq graph/activation
+footprint leaves only 9.18 GiB for the pool, and 6 × 262144 needs 12.47 GB even
+at fp8.
+
+The old "KV_BYTES is PER RANK, 9/16 GiB OOM" law was the MTP4 WORKSPACE
+reservation, not a VRAM wall. Failures are clean and diagnostic: the engine
+reports `torch.OutOfMemoryError: Tried to allocate X GiB. GPU N has ... of
+which Y GiB is free`, and the fork's giant ERROR lines are split by journald —
+read them via `journalctl -o json`, not `-o cat` (and `podman logs`, which is
+only available while the container lives).
+
+Rollback to the 2026-09-29 fp8+MTP4 shape: git tag
+`frank-primary-fp8-mtp-20261001` (env trio + seam re-pinned there).
+
+## Parallelism options on this box (2026-10-01 survey)
+
+Surveyed from the running image's own CLI surface. Verdicts are specific to
+dankserver: 2x R9700 (gfx1201), a ~172.8 GiB expert-offload MoE, and a model
+with only 2 KV heads.
+
+| option | flag | verdict on this box |
+|---|---|---|
+| TP | `--tensor-parallel-size` | **DEPLOYED, =2.** All-reduce dispatch shows `R4D_AR` then `PYNCCL` |
+| DCP | `--decode-context-parallel-size` | **REJECTED at TP=2** — see below |
+| PP | `--pipeline-parallel-size` | supported; untried (stage handoff plus a TP-keyed expert plan) |
+| DP | `--data-parallel-size` (+`-rank`, `-start-rank`) | impossible: every replica holds a full model copy (172.8 GiB vs 31.86 GiB/card) |
+| EP | `--enable-expert-parallel` + `--all2all-backend` (+`--enable-eplb`, `--enable-dbo`, `--enable-elastic-ep`) | supported (`naive`, `allgather_reducescatter`, `deepep_high_throughput`, `deepep_low_latency`, `pplx`, `mori`, `nixl`); would replace the fork's tiered-offload LOCAL gather with a per-layer PCIe all-to-all |
+| multi-node | `--nnodes`, `--node-rank`, `--master-addr/port`, `--distributed-executor-backend` | supported; the actual path to more cards |
+
+### DCP is structurally unavailable here — do not re-test on 2 GPUs
+`--decode-context-parallel-size 2` fails `VllmConfig` validation at boot:
+
+```
+Decode context parallelism for GQA/MQA requires `--tensor-parallel-size` (2) to be
+greater than the model's total number of KV heads (2). Increase
+`--tensor-parallel-size` or set `--decode-context-parallel-size 1`.
+```
+
+DCP shards the KV along the *sequence* dimension after KV heads have been spread
+across the TP ranks, so it requires **TP > num_kv_heads**. This model has 2 KV
+heads (24 Q heads, GQA 12:1), so TP must be >= 4 — four GPUs, not two. Related
+fork facts: DCP *reuses* the TP ranks rather than multiplying them (it is not
+`TP x DCP`), and with PCP enabled the valid DCP sizes are `{1, pcp, tp x pcp}`
+(`vllm/config/parallel.py`). The KV memory a DCP win would have bought is real —
+each rank would hold half the sequence's KV — but it is unreachable without two
+more cards.
+
+### Measuring decode: the harness lives INSIDE the container
+`/home/dank/mtp_measure.py` imports `/bench_iter.py` by absolute path, so it must
+run in the engine's namespace and the files must be copied in first:
+
+```
+podman cp /home/dank/mtp_measure.py frank-primary-qwen:/
+podman cp /home/dank/bench_iter.py  frank-primary-qwen:/
+podman exec frank-primary-qwen python3 /mtp_measure.py
+```
+
+Without the copy the exec dies with `can't open file '/mtp_measure.py': [Errno 2]
+No such file or directory` — measured 2026-10-01, which silently cost a whole
+arm's numbers. A window script must do the copy, not assume the path exists.
+
+### The then-deployed shape, measured again (2026-10-01 A/B window)
+A clean production-environment boot of the config deployed at that time (4 x 262144 fp8, MNBT
+2048 in that window) reported **`GPU KV cache size: 1,079,582 tokens, Maximum
+concurrency for 262,144 tokens per request: 4.12x`** — the same figures as the
+deployed MNBT-1024 shape, i.e. MNBT does not move the pool, and the capacity is
+reproducible across boots.
+
+## See Also
+
+- [[dankserver-frontier-moe-feasibility]] — the sibling feasibility page
+  (which frontier MoE fits this box; the same PCIe-expert-stream wall at
+  V4.1 scale)
+- [[frank-primary-r9v-radiance-qwen38-q8]] — RETIRED 2026-09-30 (the R9V
+  fork era; production has been the tcclaviger fork since)
+- [[x4-pair-experiment]] — the same residency law measured on a narrower
+  link: streaming over PCIe loses, computing in RAM does not care
+- [[Dual-R9700-Cooperation]] — the board swap that made TP=2 legal here

serverAppConfigs-docs/LLM/frank-large-gufo-qwen38-27b.md

@@ -0,0 +1,350 @@
+---
+type: reference
+title: "gufo, Qwen3.8-27B UD-Q4_K_XL (DFlash2) on Dankclaw (frank-large)"
+description: "The RETIRED frank-large runbook (rollback target) — gufo (gufo-org/gufo) v0.1.1, the Strix Halo inference engine, built from source on Dankclaw and serving unsloth/Qwen3.8-27B-GGUF UD-Q4_K_XL (vanilla) with the z-lab DFlash2 Q4_K_M drafter on the gfx1151 iGPU, ONE session at the full native 262,144 window with unquantized f16 KV: the from-source build and its two host workarounds, the GGUF header evidence, why the fleet's Qwen-Sharp chat template CANNOT be used on this runtime, the client-contract delta, the measured prefill/decode trade, and the failure modes worth knowing."
+tags:
+  - llm
+  - gufo
+  - swift-1.5
+  - dankclaw
+  - frank-large
+  - strix-halo
+  - gfx1151
+  - q8
+  - dflash2
+  - rocm
+timestamp: 2026-09-28
+---
+
+# frank-large — gufo, Qwen3.8-27B (UD-Q4_K_XL, DFlash2)
+
+> **RETIRED 2026-10-02 — the rollback target.** Superseded by
+> [[frank-large-rocmfpx-q8agent-mtp]] (Q8_0_ROCMFPX_AGENT at 8.39 bpw with
+> MTP4, 17.0 t/s measured). This era was itself a return after one day of
+> [[frank-large-vllm-mimo-v26-bf16]] (battery decode 6.19-9.01 t/s; every
+> ≥6-bit vLLM quant path dead on gfx1151). The shape
+> moved on the way back: payload is now **unsloth/Qwen3.8-27B-GGUF
+> UD-Q4_K_XL** (16.35 GiB, vanilla — the abliterated Swift Q8 was
+> reclaimed), and the session grid is **one session at the full native
+> 262,144 window** (`SESSIONS=1 CONTEXT_SIZE=262144`, f16 KV). The
+> Swift-1.5 Q8 history below is retained for the engine/template law,
+> which still governs.
+> **Originally ACTIVE 2026-09-28.** Replaces
+> [[frank-large-unsloth-qwen38-27b-q8xl]] (stock llama.cpp, the unsloth
+> b11030 prebuilt), which had replaced
+> [[frank-large-kairic-edge-qwen38-27b]]. Same host, same port (7770), same
+> alias (`frank-large`).
+>
+> **The payload also changed on 2026-09-28**, later the same day: the target
+> went from `unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL` to
+> `ajgazin/Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF`'s
+> `UD-Q8_K_XL` — an abliterated (uncensored) fine-tune of the same base. See
+> [Payload](#payload) and, for what that costs, [the template
+> law](#the-template-law-qwen-sharp-is-gone).
+
+## Identity
+
+- **Endpoint:** `frank-large` at `192.168.0.143:7770` (Dankclaw).
+- **Runtime:** [gufo](https://github.com/gufo-org/gufo) — "the Strix Halo
+  inference engine", a vertical engine built for exactly this hardware
+  (Ryzen AI MAX+ 395 / Radeon 8060S, `gfx1151`, unified memory). Pinned at
+  **v0.1.1** (commit `b0f8673`), built from source on the host to
+  `/home/dank/GitPackages/gufo-v0.1.1/`.
+- **Service shape:** root systemd unit `scripts/frank-large.service`
+  (`User=dank`, `LimitMEMLOCK=infinity`), launched by
+  `scripts/serve-frank-large.sh` with tunables from
+  `scripts/frank-large.env`.
+
+gufo ships **no prebuilt Linux binaries** (the v0.1.0/v0.1.1 release assets
+are empty) and no Nix on this host, so the runtime is a source build. It is
+also young: v0.1.0 landed 2026-09-28, the same day as this cutover.
+
+## Why this engine
+
+gufo's own measurement of the two engines on the same Qwen3.8-27B targets
+(Strix Halo, `628ed18e` vs llama.cpp `b11069`, greedy, DFlash2):
+
+| Q8_K_XL, DFlash2 | gufo | llama.cpp | delta |
+|---|---:|---:|---:|
+| prefill, depth 0 | 457.05 t/s | 333.85 t/s | **+36.9 %** |
+| prefill, depth 131,072 | 237.24 t/s | 131.80 t/s | **+80.0 %** |
+| decode, depth 0 (mixed) | 16.08 t/s | 15.00 t/s | +7.2 % |
+| decode, depth 32,768 (mixed) | 14.41 t/s | 14.44 t/s | −0.2 % |
+| decode, depth 65,536 (mixed) | 10.87 t/s | 12.62 t/s | **−13.9 %** |
+| decode, depth 131,072 (mixed) | 8.87 t/s | 11.62 t/s | **−23.7 %** |
+| cold load to HTTP-ready | 8.17 s | 36.47 s | **4.5× faster** |
+| peak memory @ 262,144 ctx | 50.58 GiB | 48.60 GiB | +4 % |
+
+Read that honestly: **prefill, load time and short-context decode improve;
+deep-context decode on the 8-bit target gets worse.** The crossover on Q8 is
+around 32k of cached depth. frank-large runs one 262,144-token conversation,
+so a long agent session spends its life on the losing side of that line
+while a short turn wins. The Q4_K_XL target does not show the regression
+(gufo is +11.7 % to +17.0 % at every depth) — that is the honest reason to
+revisit the quantization separately from the runtime.
+
+## Build
+
+`scripts/build-gufo.sh` reproduces the runtime end to end. It clones the
+pinned tag, configures the `release` preset (HIP, `--offload-arch=gfx1151`),
+builds, installs, and finishes with `gufo diagnose`. Two host facts it
+encodes, both found the hard way on 2026-09-28:
+
+- **Ubuntu's `librocwmma-dev` must not be installed.** It ships a truncated
+  rocWMMA header set (the public headers without `rocwmma/internal/`) into
+  `/usr/include`, where it shadows AMD's complete copy under
+  `/opt/rocm/include`. The first build died on `'internal/accessors.hpp'
+  file not found` from `/usr/include/rocwmma/rocwmma.hpp`, and **a `-I`
+  cannot fix it**: in HIP mode the host compilation pass searches the system
+  directories first. The package has no reverse dependencies; the script
+  removes it.
+- **`--gcc-install-dir` must point at GCC 15.** Ubuntu 26.04 has
+  `/usr/include/c++/16`, and the ROCm clang picks the newest tree, whose
+  `<atomic_wait.h>` references `std::__detail::__notify_impl` at a symbol
+  version the installed `libstdc++.so.6` (GLIBCXX_3.4.35) does not export.
+  The link then fails with `undefined symbol`. Pinning the headers to the
+  GCC the runtime library came from fixes it.
+
+Verify an installed runtime with `bash scripts/build-gufo.sh --verify`, or
+`gufo diagnose` directly — it reports `[PASS]` plus the detected
+architecture, driver and queue budget.
+
+## Shape
+
+### Payload
+
+- **Model:** `ajgazin/Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF`,
+  file `UD-Q8_K_XL` — 27B dense, 31,457,990,272 bytes, fetched and
+  sha256-verified (`9076f99a…`, the Hub LFS oid) by
+  `scripts/download-frank-large-swift.sh`. It is an **abliterated
+  (uncensored)** fine-tune of Qwen3.8-27B. The repository also publishes an
+  `mmproj-BF16.gguf`, which is **not** used: this endpoint is text-only.
+- **Licence:** the repo declares `other` — the **SWIFT OPEN LICENSE 1.0**,
+  which is *not* an OSI licence. That is a real constraint on redistribution,
+  noted here so nobody discovers it later. It is fine for a private fleet.
+- **Replaced:** `unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL`, still on disk and
+  still the rollback target for the payload.
+
+The two targets are **structurally interchangeable for this engine**, which
+is why the swap needed no engine change. Header-diffed directly:
+
+| | unsloth | Swift-1.5 |
+|---|---|---|
+| `general.architecture` | `qwen35` | `qwen35` |
+| tensors (count, names, order) | 866 | 866, **identical** |
+| per-tensor `ggml_type` | Q8_0×453, F32×360, BF16×53 | **identical** |
+| per-tensor dims | — | **0 mismatches** |
+| MTP head | 4 tensors at `blk.64.nextn.*` | identical |
+| geometry | 65 blk, 5120 embd, 17408 FFN, 24/4 heads | identical |
+
+The two metadata differences that look alarming are both no-ops, and both
+were confirmed against gufo's source rather than assumed:
+
+- **`general.file_type` is 15 vs 7.** gufo's `gguf_reader.cpp` states it is
+  *"deliberately not consulted… The Unsloth UD-\*_K_XL artifacts this repo
+  targets carry a stale one — UD-Q8_K_XL records ftype 15 (Q4_K_M) … The
+  tensor table always is."* The file gufo already served lies about itself
+  here, so gufo reads the tensor table.
+- **Swift adds `qwen35.attention.recurrent_layers`.** gufo never reads that
+  key anywhere in `src/`; layer kinds come from `full_attention_interval`,
+  which both files carry as `4` and which agrees with it.
+
+### Serving shape
+
+- **Context:** one session, full 262,144 tokens (the model's native window),
+  `SESSIONS=1`. The fleet's single stream is unchanged.
+- **Spec method:** DFlash2, gufo's **adaptive** block policy
+  (`--draft-policy adaptive`, `--draft-tokens 7`). The drafter is the
+  **unchanged** z-lab `Qwen3.8-27B-DFlash2-Q4_K_M` (1.07 GiB) — it did not
+  move with the target, which is why `DRAFTER_DIR` is a separate tunable
+  from `MODEL_DIR` and why the deployer's stray-run sweep keys on the
+  drafter's filename rather than the target's.
+- **Sampling:** greedy (`--temperature 0`) server-side, as every prior
+  frank-large served. The harness overrides per request via `extraBody`
+  (1.0 / 0.95 / top-k 20).
+- **Memory:** gufo's own table puts the 8-bit target at **50.58 GiB** peak
+  with context capacity 262,144, against the iGPU's ~114 GiB GTT aperture.
+
+### Measured at the 2026-09-28 cutover
+
+Live on Dankclaw. The **target** numbers are the Swift-1.5 load test
+(a standalone run under the production flags, service stopped first); the
+**engine** numbers are from the gufo cutover earlier the same day and are
+unchanged by the payload swap.
+
+- **9.2 s** `load_started` → `load_completed` for the Swift target, against a
+  20-minute `LLM_WAIT` budget. The unsloth target loaded in 13 s.
+- **66.5 GiB** of the 116.7 GiB iGPU device at context capacity 262,144;
+  host RSS 30.6 GiB. No loader warnings, no skipped sections.
+- Contract battery, all green on the Swift target: `/v1/models` →
+  `context_length 262144`, a chat completion carrying both `content` and
+  `reasoning_content`, a `tool_calls` response with
+  `finish_reason: tool_calls`, SSE streaming, and a **two-turn tool loop**
+  (tool call → tool result → follow-up) with prior reasoning preserved.
+  Arithmetic spot-check: "17*23" → `391`.
+- Behavioural spot-check that this is the abliterated model: asked for a
+  two-sentence lockpicking scene, it wrote one without hedging or refusing.
+- From the gufo cutover, on the unsloth target: a short-prompt live request
+  returned
+  `Server-Timing: ttft;dur=549.8, inter_token;dur=24.2` (≈41 tok/s).
+  Short-prompt decode is gufo's *winning* regime; see the depth table above
+  before quoting it as the endpoint's number.
+
+### Tuning: the served flags are already optimal (2026-09-28)
+
+Frank asked for iterative decode/prefill improvement at a 32k working point
+without touching the model or the quantization. **Result: no win exists at
+this engine's served settings, and nothing was changed.** Measured on real
+repo prose and shell sources at **32,762 prompt tokens**, salted per run so
+a warm prefix cache could not help, 3–5 repetitions each.
+
+| lever | range swept | outcome |
+|---|---|---|
+| `--prefill-chunk` | 512 → 8192 | **no effect** — 410.7–412.4 t/s prefill across a 16× change |
+| `--draft-tokens` | 5 → 10 | **flat** — 13.79–15.86 t/s decode, inside run-to-run noise |
+| `--draft-policy` | fixed vs adaptive | 35.90 vs 35.50 (bench) — no effect |
+| `--min-draft-tokens` | — | **rejected by the engine**: "DFlash2 requires … 1" |
+| MTP backend | — | **unsupported for Q8**: `token_embd.weight` is Q8_0 |
+| `--context` 262144 → 40960 | diagnostic | 13.93 vs 14.71 t/s — no gain, and it would break the declared client context |
+
+The confirmation round at 5 reps **reversed** the 3-rep leaders: the pair
+that looked best (`dt9 + chunk4096`) came last at 12.86 t/s, and the
+**deployed configuration came first at 14.71 t/s**. Every difference sits
+inside the noise band.
+
+What *does* move the number is speculative decoding itself: **6.65 t/s off →
+~15 t/s on, a 5.3× step** — and that is already switched on. Prefill is
+insensitive to everything tried, which is consistent with gufo's own
+`EXPERIMENTS.md`, where dozens of kernel-level alternatives are recorded as
+rejected for "no release throughput improvement".
+
+**Do not repeat this investigation without new information.** Two traps
+cost most of the time, and both are in AGENTS.md's benchmarking policy:
+`gufo bench` reports `acceptance=0.933 ± 0.000` on an identical trace hash
+and **inverts** the draft-width ranking; and a live A/B needs ≥128 generated
+tokens or `inter_token` is first-token noise.
+
+## The template law: Qwen-Sharp is gone
+
+`scripts/qwen-sharp-v22.4.0.jinja` is the fleet's chat template — it is
+wired into `frank-primary` as `R9V_CHAT_TEMPLATE_FILE` and it was
+`frank-large`'s `--chat-template-file` under llama.cpp. **frank-large can no
+longer use it**, and the reason is stronger than "gufo lacks the flag":
+
+> gufo has **no Jinja engine at all.** `src/models/qwen/chat_template.cpp`
+> is 587 lines of hand-written C++ that appends strings to build ChatML.
+> There is no `--chat-template-file`, no environment override, and no
+> template body anywhere in the binary to point at. Qwen-Sharp is 453 lines
+> of Jinja2 using `namespace()`, macros, `raise_exception` and image/video
+> branches — supplying it would mean **reimplementing it in C++**, not
+> handing gufo a file.
+
+Given the choice on 2026-09-28, the decision was to **ship on gufo's stock
+Jinja and accept the loss**. What that gives up, concretely:
+
+| Qwen-Sharp feature | On gufo |
+|---|---|
+| `terse` system-prompt injection (the headline behaviour) | **gone** |
+| in-band `<\|think_off\|>` / `<\|think_xhigh\|>` control | **gone** |
+| `tool_call_format` `json` \| `xml` switch | **gone** |
+| `max_tool_arg_chars` / `max_tool_response_chars` truncation | **gone** |
+| `add_vision_id` → `Picture N:` prefixes | **gone** |
+| effort instruction strings in the system prompt | kept (gufo has its own, same text) |
+| `preserve_thinking` | kept (`--preserve-thinking on`) |
+
+The `terse` injection is the one that matters most here, and it is worth
+being blunt about the combination: this endpoint now runs an **abliterated**
+model *without* the prompt that tells it to stop waffling. If verbosity
+shows up, the cheapest fix is to send the `terse` text as an ordinary
+**system message from the client** (`models.yml` / the harness) rather than
+through a template — that recovers the behaviour without forking the
+renderer. The full-fidelity options are reverting to llama.cpp or porting
+Qwen-Sharp into the C++ renderer; both are worse trades today, and the
+second is the same class of change that produced the 2026-09-21
+"73,600-token prompt" incident.
+
+## Client contract — what actually changed
+
+Everything the fleet depended on still holds: port 7770, alias `frank-large`,
+`reasoning_content`, `tool_calls`, streaming SSE, and text-only (gufo
+auto-discovers a `mmproj` beside the model; this payload has none).
+
+The differences worth knowing:
+
+- **No custom chat template — see
+  [The template law](#the-template-law-qwen-sharp-is-gone) above**, which is
+  the most consequential change on this endpoint. Thinking defaults to on at
+  `xhigh` with prior reasoning preserved (`--think auto
+  --preserve-thinking on`, spelled out in the serve script).
+- **`reasoning_effort` accepts the full OpenAI set** (`none`, `minimal`,
+  `low`, `medium`, `high`, `xhigh`, `max`), all verified 200 against the
+  live endpoint. Qwen-Sharp rejected anything outside `{xhigh, medium, low}`
+  with an HTTP 500, which is why `models.yml` still carries a
+  `reasoningEffortMap` fold for this model. **The fold is kept on
+  purpose**: removing it would silently move every `high` role from `xhigh`
+  to `high` effort. Drop it as its own change.
+- **Every field the harness sends is accepted.** `top_k`, `min_p`,
+  `presence_penalty`, `repetition_penalty` and `preserve_thinking` are all
+  outside gufo's documented request subset, and all returned 200. They are
+  applied from `models.yml`'s `extraBody` unchanged.
+- **No KV-cache quantization flag.** gufo picks its own KV dtype; there is
+  no `--cache-type-k`. The old `CACHE_RAM=8192` tunable is gone with it.
+  Prompt reuse is on by default, and `--cache-disk` (restart-safe) is opt-in
+  and currently off.
+- **Health/readiness split:** `GET /health` is process liveness and is what
+  the deployer gates on; `GET /ready` is model/backend readiness.
+  `GET /metrics` is Prometheus-format.
+
+- **Long bodies:** `--max-request-bytes` defaults to 8 MiB.
+
+## Failure modes worth knowing
+
+- **A missing runtime is a deploy error, not a restart loop.** The handler
+  gates on `LLM_LARGE_RUNTIME` *before* it stops anything, and the serve
+  script exits with a `build-gufo.sh` pointer rather than exec'ing.
+- **Host memory is the ceiling, and it is shared.** frank-large and ComfyUI
+  both hold gfx1151 compute queues. gufo reads
+  `/sys/class/kfd/kfd/proc/*/queues/*/type` and caps itself
+  (`GPU_MAX_HW_QUEUES`, 2 resident for `serve llm`) so a full device
+  time-slices instead of failing; four HIP processes is the device ceiling.
+- **Load is seconds, not minutes** (9.2 s measured cold on the Swift
+  target), so a redeploy no longer spends 20 minutes in `LLM_WAIT`.
+- **A payload is two files in two directories.** The target and the drafter
+  no longer live together, so "the payload is present" is not one check; the
+  deployer gates on both paths and the serve script refuses to start if
+  either is absent. A symlinked "one payload directory" would hide that.
+- **A v0.1.x runtime is a moving target.** Rebuilds are not automatic —
+  `RUNTIME_DIR` names a version, and the deployer will not silently
+  substitute another.
+
+## Rollback
+
+Two independent rollbacks, because two things changed on 2026-09-28:
+
+1. **Payload only** — repoint `MODEL_FILE` in `scripts/frank-large.env` at
+   `/home/dank/models/Qwen3.8-27B-UD-Q8_K_XL/Qwen3.8-27B-UD-Q8_K_XL.gguf`
+   and redeploy. **The 29.3 GiB target was reclaimed on 2026-09-28**, so this
+   path now needs a re-fetch first: `scripts/download-frank-large.sh`, then
+   the env edit. The DFlash2 drafter was never touched and still lives in
+   that directory.
+2. **Engine** — [[frank-large-unsloth-qwen38-27b-q8xl]]: the unsloth b11030
+   prebuilt is still at `/home/dank/GitPackages/unsloth-llama-b11030/`.
+   Revert `RUNTIME_DIR` and the gufo-only tunables, restore the serve
+   script's llama.cpp command, and redeploy. This is also the only way to
+   get `qwen-sharp-v22.4.0.jinja` back.
+
+# Citations
+
+- Runtime: https://github.com/gufo-org/gufo (pinned `v0.1.1`, MIT)
+- Measured tables: `docs/models/qwen3.8-27b/BENCHMARKS.md` in that repo
+- API contract: `docs/SERVER.md`; flags: `docs/CLI.md`; the C++ renderer
+  that rules out a Jinja template:
+  `src/models/qwen/chat_template.cpp` (587 lines)
+- GGUF header evidence for the payload swap: both files' headers were parsed
+  and diffed directly (architecture, all 866 tensor names in order, per-tensor
+  `ggml_type` and dims) before either was loaded
+- Build: `scripts/build-gufo.sh`; serve: `scripts/serve-frank-large.sh`;
+  tunables: `scripts/frank-large.env`
+- Payload: `scripts/download-frank-large-swift.sh` (target),
+  `scripts/download-frank-large.sh` (DFlash2 drafter)
+- Predecessor: [[frank-large-unsloth-qwen38-27b-q8xl]]

serverAppConfigs-docs/LLM/frank-large-halogen27b-dflash2.md

@@ -0,0 +1,130 @@
+# frank-large — halogen container, Qwen3.8-27B + DFlash2 (Dankclaw, ACTIVE 2026-10-05)
+
+frank-large is the Strix Halo iGPU endpoint on Dankclaw (`:7770`). Since
+2026-10-05 it runs the **halogen** engine — `ghcr.io/peonist-ai/halogen` at a
+pinned digest, launched by `scripts/serve-frank-large.sh` under a root
+systemd unit as a podman container — serving
+**`peonist-ai/halogen-qwen3.8-27b`** (`qwen3.8-27b-p1w4d-d2.hgn`,
+35,865,565,184 B ≈ 33.4 GiB, sha256 `274c3fc7…`, 1352 tensors) with the
+**DFlash2** drafter at **one slot x 200,700 ctx**.
+
+It replaced the ROCmFPX llama.cpp fork endpoint ([[frank-large-rocmfpx-q6agent-uncensored-mtp]]),
+which remains the rollback target with its payload on disk.
+
+## Why swap
+
+|Metric|ROCmFPX Q6_0_ROCMFPX_AGENT + MTP (retired)|halogen qwen3.8-27b + DFlash2 (active)|
+|---|---|---|
+|Prose decode|27.6–28.1 t/s measured (4.8k depth, acceptance 0.868)|31.7 t/s mean across eight text workloads (publisher card, bf16 reference hardware)|
+|Copy-echo decode|31.4 t/s|publisher reports echo inside the same 31.7 mean|
+|Context|1 x 262,144|1 x 200,700|
+|Format|GGUF (fork-private tensor types 102/103)|`.hgn` — loads in NO other engine|
+
+The swap was accepted with eyes open on two product regressions, both
+confirmed by Frank before deploy:
+
+1. **Text-only.** halogen loads the language tower only; README "honest
+   limits": *"The vision encoder is ignored; image inputs are not supported."*
+   The mmproj-era vision role moves wholly to frank-small.
+2. **Censored base.** The checkpoint is a finetune of the censored
+   `Qwen/Qwen3.8-27B` base, built to make its DFlash2 head strong. The
+   ROCmFPX AGENT file was an *Uncensored* build; refusals can return.
+
+## Shape and why
+
+`scripts/frank-large.env` carries every tunable with its evidence; the load-
+bearing numbers:
+
+- **`HALOGEN_KV_SLOTS=1` + `HALOGEN_DRAFTER=2`** — the engine offers batch OR
+  speculation, never both (FLAGS.md: "Batching and speculation are exclusive:
+  a step batches when more than one conversation is running, so the drafter
+  stays off with more than one slot"). Slots>1 silently loses DFlash2; the
+  sibling halogen engine measured 41 -> 14 t/s when slots went up. frank-large
+  was a one-conversation endpoint before (PARALLEL=1), so the contract holds.
+- **`HALOGEN_SLOT_CTX=200700`** — the per-conversation HARD ceiling: a longer
+  prompt is a 400 naming the limit, never a truncation. The harness side is
+  wired to it: `models.yml` `contextWindow: 200700`, and OMP compacts at 70%
+  = 140,490, leaving 60,210 of the slot for the reply.
+- **`HALOGEN_MAX_TOKENS_CAP=65536` + `HALOGEN_QUEUE_TIMEOUT=7200`** — coupled
+  pair (FLAGS.md table): a request at the cap must not outlive the timeout.
+  65536 > the harness's 57,344 `maxTokens` so an eager request never meets
+  the engine's 400.
+- **KV math:** 1 slot x 200,700 x 64 KiB/token (d2 KV = 2x fp8) = 12.2 GiB
+  pool, inside the ~65 GiB left after 33.4 GiB mmap'd weights + engine working
+  set + host baseline on the 121.2 GiB box. The engine sizes the pool itself
+  and **prints a boot ledger naming any clamp** — the post-deploy check reads
+  that ledger, not a formula.
+- **No server-side sampler:** an absent temperature is GREEDY. The card's
+  thinking recipe (temp 1.0 / top_p 0.95) lives in `models.yml` `extraBody`
+  — on the wire only. The template defaults `reasoning_effort` to **xhigh**
+  when a request sends none, and **thinking tokens consume the
+  `max_tokens` budget**: a budget exhausted mid-thought returns
+  `finish_reason: "length"` with EMPTY `content` and the partial reasoning
+  in `reasoning_content` — clients must read both fields.
+
+## Client contract deltas vs llama.cpp
+
+|Behavior|halogen|
+|---|---|
+|Prompt over slot|400 naming the limit (no truncation)|
+|`max_tokens` over cap|400 naming the limit (never silent truncation)|
+|full-length request outliving `HALOGEN_QUEUE_TIMEOUT`|503s for everyone queued|
+|`timings`|every response carries llama-server-shape `timings` (`draft_n`, `draft_n_accepted`) — the drafter check reads the RESPONSE, and it proves speculation fired|
+|`response_format`|NOT supported (no constrained decoding)|
+|thinking field|`reasoning_content` (same as before)|
+|`reasoning_effort`|same Qwen3.8 template; outside {xhigh,medium,low} rejected — models.yml fold stays; unset defaults to xhigh, `none` disables thinking|
+|tools|server-side tool calls advertised — round-trip MUST be re-probed after deploy|
+|image input|ignored/error — route to frank-small|
+
+## Deploy mechanics
+
+- **Image:** `ghcr.io/peonist-ai/halogen@sha256:dc0a39a0016d6cfc58a197978febaafdf8d28403f724ded6111d98b5fb7ac0ea`
+  (published tag 0.1.4). The deployer gates the exact digest exists in podman
+  storage before dispatch. NOT the 2026-09 `halogen-flash-server` repo — that
+  was the Flash-Next w4b era, different image, retired.
+- **Payload:** `scripts/download-frank-large-halogen.sh` (fetch law: `curl -f`,
+  exact byte count, LFS sha256 where the Hub publishes one, `--verify-only`,
+  token never in argv) into `/home/dank/models/halogen-qwen3.8-27b/`. The
+  tokenizer is a FLAT DIRECTORY (`tokenizer/`, 5 files — tokenizer.json,
+  vocab.json, merges.txt, chat_template.jinja, tokenizer_config.json),
+  mounted whole at `/tokenizer`; the engine's `HALOGEN_TOKENIZER` default is
+  that directory, so the container gets no tokenizer flag (model-card
+  docker-run parity). The deployer and serve script refuse a missing file or
+  ANY symlink: an HF cache snapshot symlinks each entry into `blobs/` and
+  those links dangle inside the container mount.
+- **Unit:** `scripts/frank-large.service` (root, `User=dank`, relinked into
+  `/etc/systemd/system/`), `ExecStart` = serve script, `ExecStop` = `podman
+  stop -t 120 frank-large-halogen`, script `exec podman wait` so
+  `Restart=always` is real. Engine stdout goes to
+  `/home/dank/logs/frank_large.log` (the boot ledger lives there, not in the
+  journal).
+- **Cutover:** `deploy_llm_large` in `scripts/deploy-remote.sh` sweeps the
+  legacy bare `llama-server --model /home/dank/models/kingjones-*` process and
+  the legacy `frank-large` container name by NAME (never `pkill -f
+  llama-server` — that hits frank-small), retires the old quadlet if the 2026-09
+  halogen era's unit lingers, waits port 7770 free, relinks, health-gates.
+- **Bench law:** stop+mask the unit before any GPU experiment on this box
+  (AGENTS.md); `/health` answers 503 while the 33.4 GiB payload loads — that
+  is loading, not failure.
+
+## Verification (post-deploy checklist)
+
+1. `grep -iE 'clamp|kv|pool|drafter|dflash' /home/dank/logs/frank_large.log`
+   — ledger shows the pool at the asked shape, DFlash2 active, NO clamp line.
+2. `curl -s http://192.168.0.143:7770/health` -> 200.
+3. Real completion with `reasoning_effort` + temp/top_p from the wire recipe;
+   `reasoning_content` populated; decode rate from the timings.
+4. Spec decode fires: multi-token step behavior visible in rate; compare
+   against DRAFTER=0 arm if in doubt (never against gufo — bench law).
+5. Harness round-trip: an OMP call through `llamacpp (dankclaw)/frank-large`
+   incl. a tool call (V1TOOLS).
+6. Prompt > 200,700 tokens returns the 400 naming the limit (contract test).
+
+## Rollback
+
+Revert the frank-large swap commit and redeploy `dankclaw --service frank-large`:
+the ROCmFPX payload (`/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP/`),
+the fork build (`/home/dank/GitPackages/kingjones-ROCmFPX/build`), and its
+`scripts/build-rocmfpx.sh` pin (commit `36e9acd`, seam-tested) all remain on
+disk. The halogen payload can stay: it costs nothing while the unit points
+elsewhere.

serverAppConfigs-docs/LLM/frank-large-ik-uncensored.md

@@ -123,3 +123,27 @@ Stop and disable the root unit, trash it, restore the halogen quadlet from
 `~/.AGENTTRASH/20260915-ik-swap/halogen-quadlet` to
 `~/.config/containers/systemd/`, `systemctl --user daemon-reload`, start;
 the .hgn checkpoint and the kingjones STRIX llama.cpp payload remain on disk.
+
+## Hybrid campaign envelope (2026-09-16, 3090 + Strix Halo)
+
+Why this deployment could not be tuned into a production endpoint: decode
+3.8–4.3 tok/s flat across thread counts (T1 ≈ T16); prefill ~192 tok/s at
+1.8k–32k prompts = the **CPU AVX512-VNNI compute ceiling** for CPU experts.
+The decode root cause is structural: ~100 CPU↔GPU backend splits per token
+with the GPU at 5 % SM during decode — launch/sync-bound over the PCIe x4
+dock, not compute- or memory-bound. T16 profiling shows ~90 % of cycles in
+`pause` (ggml_barrier spin): threads wait on the serial critical path.
+Memory placement (mmap/anon/THP/deferred) does not matter — all variants
+landed 3.8–4.6.
+
+- CUDA-graph capture aborts: qwen4exp's hyper-connection broadcasts use
+  per-layer `ggml_repeat_4d`, whose CUDA path is capture-illegal; until ik
+  gains capture-safe repeat/hc ops (mainline PR #28901 direction), graphs
+  stay dead.
+- Mainline llama.cpp itself runs 1.0 tok/s on this model (CPU MoE kernels
+  4× slower than ik's iqk) — do not switch engines for speed.
+- MTP/speculative decoding is impossible for mradermacher i1 GGUFs: the
+  NextN head is stripped (no `blk.48` tensors).
+- Tuning lever ranked first: `-ncmoe 42` (6 layers' experts on the 3090) →
+  decode 4.54, small-prompt prefill 32 — but VRAM 22/24.6 GiB at 262k ctx,
+  an operator OOM-margin decision, and it needs `GGML_CUDA_NO_PINNED_WEIGHTS=1`.

serverAppConfigs-docs/LLM/frank-large-rocmfpx-q6agent-uncensored-mtp.md

@@ -0,0 +1,177 @@
+---
+type: reference
+title: "frank-large on Dankclaw — ROCmFPX llama.cpp, Qwen3.8-27B Uncensored Q6_0_ROCMFPX_AGENT + MTP"
+description: "The active frank-large runbook — the kingjones30/ROCmFPX llama.cpp fork (pinned commit, gfx1151) serving kingjones777's Uncensored Q6_0_ROCMFPX_AGENT GGUF (ftype 114, protected output heads), with the repo's own MTP-Q4_0 head drafting four tokens at one native 262,144 window, f16 KV, the BF16 projector for vision, and the measured rejection of the ngram-mod chain."
+tags:
+  - llm
+  - frank-large
+  - dankclaw
+  - llama-cpp
+  - rocmfpx
+  - gfx1151
+  - gguf
+  - spec-decode
+  - mtp
+timestamp: 2026-10-03
+---
+
+# frank-large — ROCmFPX Uncensored Q6_0_ROCMFPX_AGENT + MTP (gfx1151)
+
+> **RETIRED 2026-10-03 → 2026-10-05 (rollback target; payload, fork build and
+> build script all on disk).** Replaced by [[frank-large-halogen27b-dflash2]]
+> (DFlash2, publisher 31.7 t/s mean vs the 27.6–28.1 measured here). This page
+> remains the contract record for the rollback. Original banner follows.
+> ACTIVE since 2026-10-03. Cut over from
+> [[frank-large-rocmfpx-q8agent-mtp]] (ALIGNED Q8_0_ROCMFPX_AGENT, 17.0 t/s)
+> on Frank's request: the *uncensored* variant of the same protected-heads
+> family, one quant rung down. The swap turned out to be a strict upgrade on
+> the axis that matters here: **decode 27.6–28.1 t/s at a 4,813-token prompt
+> against 17.0 at a 2,202-token prompt** — faster with a deeper context, not a
+> same-depth comparison, but the sign is not in question.
+
+## Identity
+
+| | |
+|---|---|
+| Host / device | Dankclaw (192.168.0.143), Radeon 8060S iGPU `gfx1151`, ~114 GiB GTT aperture |
+| Endpoint | `http://192.168.0.143:7770` — alias `frank-large`, `/v1` OpenAI surface |
+| Runtime | **`kingjones30/ROCmFPX`** llama.cpp fork @ `36e9acd40e10a87cd3c3ef8ec734668757dc8520`, built by [`scripts/build-rocmfpx.sh`](../../scripts/build-rocmfpx.sh) — unchanged from the Q8 era; the fork already knew type 102 |
+| Unit | root systemd `frank-large.service` ([`scripts/frank-large.service`](../../scripts/frank-large.service)), `User=dank`, `Restart=always` |
+| Serve script | [`scripts/serve-frank-large.sh`](../../scripts/serve-frank-large.sh) (owns the argv), tunables [`scripts/frank-large.env`](../../scripts/frank-large.env) |
+| Weights | `/home/dank/models/kingjones-Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP/Qwen3.8-27B-Uncensored-Q6_0_ROCMFPX_AGENT.gguf` — 24,856,607,008 B, sha256 `e966deb3…` |
+| Draft | `…/mtp-Qwen3.8-27B-Q4_0.gguf` — 1,680,271,648 B, sha256 `051a1764…` (byte-identical to the aligned repo's head; hardlinked, not re-downloaded) |
+| Projector | `…/mmproj-Qwen3.8-27B-BF16.gguf` — 931,145,888 B (hardlinked from the aligned payload dir; the same file the Q8 era served) |
+| Fetch | [`scripts/download-frank-large-kingjones-uncensored.sh`](../../scripts/download-frank-large-kingjones-uncensored.sh) — hardlinks the draft/projector from the aligned dir when present, otherwise fetches; `-f`, byte-count + LFS sha256 verified, `--verify-only`, token never on the command line |
+| Seam | `tests/frank_large_unit.bats` (payload gates, deployer/serve parity, spec block, KV dtype) |
+
+## Why this file
+
+`kingjones777/Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP-GGUF` ships the same
+"protected heads" construction as the aligned repo, in the uncensored tuning.
+Parsed header of the deployed file:
+
+| Field | Value |
+|---|---|
+| ftype | **114** — the fork's 64-bit-block family, ≈7.44 bpw |
+| Tensor mix | 353 F32 + **339 type-102 (`GGML_TYPE_Q6_0_ROCMFPX`)** + **159 type-103 (`Q8_0_ROCMFPX`)** |
+| Protected heads | 64 output-head tensors kept at plain precision — the AGENT construction from the aligned repo, at the 6-bit rung |
+| Arch | qwen35, 64 layers, `full_attention_interval 4`, `context_length 262144` |
+| Embedded sampling recipe | temp 1.0 / top_p 0.95 / top_k 20 |
+
+The same lock-in law as the Q8 era applies: types 102 and 103 exist in no
+loader but this fork; stock llama.cpp and gufo refuse the file at load
+(`unknown tensor type`). The loader was not re-litigated — the file picks the
+runtime, and the pin in `scripts/build-rocmfpx.sh` (commit by SHA) stays.
+
+Rejected sibling: `Q6_0_ROCMFPX_LEAN` (6.45 bpw, no protected heads) — the
+protection is what keeps the output-head logits at full precision and the
+drafter accurate, so it stays deployed.
+
+## Shape and budget
+
+Unchanged from the Q8 era — the swap was payload-only:
+
+| Dial | Value | Why |
+|---|---|---|
+| `--ctx-size` | 262,144 | the native window, no rope scaling |
+| `--parallel` | 1 | one conversation, all memory to it |
+| KV dtype | **f16** (both K and V) | ~64 KiB/token; free inside a 114 GiB aperture |
+| `--fit off` | on | the auto-fitter would silently resize the window |
+| `--flash-attn on` | on | required for the long-window attention path |
+| `--direct-io`, `--ctx-checkpoints 16`, `--cache-ram 8192`, `--cache-idle-slots`, `--jinja`, `--reasoning on --reasoning-format deepseek --reasoning-budget -1` | | same as [[frank-large-rocmfpx-q8agent-mtp]] |
+
+Spec decode: `--spec-type draft-mtp`, the repo's `mtp-Qwen3.8-27B-Q4_0.gguf`,
+`--spec-draft-n-max 4`, pinned `--spec-draft-device ROCm0`.
+
+Measured steady state: GTT **~61 GiB** under live load (host baseline 16.7),
+boot to `/health` **~15 s**.
+
+## Measured on this payload (2026-10-03, production service, `/completion` timings, temp 0, `cache_prompt: false`)
+
+| Traffic | Decode t/s | Prefill t/s | Draft acceptance |
+|---|---|---|---|
+| Prose, 4,813-token prompt, 512 tokens | **27.57 / 28.10** (two runs) | ~205–217 | **396/456 = 0.868** → mean acceptance length ≈ 4.4, i.e. drafting at nearly the full n-max 4 |
+| Schema echo (structured repetition) | **31.44** | — | 639/639 = 1.000 |
+| Block copy (text repeated verbatim) | ~28.0 | — | 0.848 |
+
+vs the Q8 era's 0.462 acceptance at a 2,202-token prompt: the MTP head drafts
+materially better on this file. The card's claimed 0.953-class acceptance is
+no longer science fiction on real prose — but keep reading it as a file
+property, not a throughput promise.
+
+## The ngram-mod chain — measured and rejected
+
+Frank's ask: layer `--spec-type draft-mtp,ngram-mod` on top of MTP for an
+extra gain. Measured in three bench windows (production service stopped per
+the benchmarking law; artifacts `/home/dank/fl-ngram-ab/` on Dankclaw),
+comparing `draft-mtp` alone vs the chain on the same traffic:
+
+| Traffic | MTP alone | MTP + ngram-mod | Delta |
+|---|---|---|---|
+| Prose | 27.57 t/s | 23.40 t/s | **−15 %** |
+| Block copy | 28.0 t/s | 26.1 t/s | **−7 %** (with ngram-mod demonstrably firing: 138 proposals/run) |
+
+Mechanism, straight from arm B's boot log:
+
+```
+recurrent state rollback is not compatible with 'ngram-mod' - disabling rollback support
+```
+
+On the qwen35 hybrid (Gated DeltaNet recurrent layers) adding the ngram layer
+disables the rollback path MTP's own verification relies on, so acceptance —
+and throughput — drop on *all* traffic, even where the ngram matcher fires.
+**Deployed decision: `draft-mtp` alone.** Chaining is not free on recurrent
+models; the fleet law in `AGENTS.md → llama.cpp serving` records it.
+
+## Bench-window hardening laws (learned building the A/B)
+
+1. **Health 200 is not proof of ownership.** After boot, verify the pid you
+   started owns the port (`ss -ltnp`); a zombie server from a prior window
+   answers `/health` too and silently serves your arm.
+2. **Kill scratch servers with SIGKILL after a grace period.** llama-server's
+   graceful SIGTERM waits on open keep-alive connections — a harness holding
+   the socket leaves the process (and its ~24 GiB of GTT) alive after the
+   script exits.
+3. **Prove the port is free and GTT dropped between arms**, or the second
+   arm measures a collision.
+
+## Client contract
+
+Unchanged from the Q8 era and re-verified on this payload 2026-10-03: model id
+`frank-large`, native tools, thinking in `reasoning_content`,
+`reasoning_effort` ∈ {xhigh, medium, low} only (anything else HTTP 500),
+card-recipe sampling defaults server-side, vision through the BF16 projector
+(red PNG → **`Red`**, 74 prompt / 165 completion tokens). The traps list —
+`content: null` when reasoning eats the budget, the self-recovering MTP
+boundary log line, `enable_thinking:false` not `reasoning_effort:"none"` — is
+in [[frank-large-rocmfpx-q8agent-mtp]] and still applies.
+
+Sweep key for scratch servers is now the payload name:
+`pkill -f 'Qwen3.8-27B-Uncensored-Q6_0_ROCMFPX_AGENT'` — never `pkill -f
+llama-server` (frank-small is also one).
+
+## Operations
+
+```bash
+./scripts/deploy.sh dankclaw --service frank-large
+systemctl status frank-large
+journalctl -u frank-large -n 200 --no-pager
+scripts/download-frank-large-kingjones-uncensored.sh --verify-only
+```
+
+## Rollback
+
+| Target | How |
+|---|---|
+| **ALIGNED Q8_0_ROCMFPX_AGENT** (previous shape, 17 t/s) | the payload dir `/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP/` is untouched on disk; revert the swap commit (unit/env/serve trio) and redeploy, or point `FRANK_LARGE_MODEL_FILE` + `LLM_LARGE_MODEL_FILE` at the old paths. Verify with `scripts/download-frank-large-kingjones.sh --verify-only` |
+| gufo Q4_K_XL era | git tag `frank-large-gufo-q4kxl-20261001`, **but the gufo payloads (UD-Q4_K_XL 16.35 GiB, UD-Q8_K_XL + DFlash2 drafter) were removed from disk 2026-10-03 with Frank's approval** — this rollback now requires re-download per [[frank-large-gufo-qwen38-27b]] |
+
+## When it misbehaves
+
+Same decision tree as [[frank-large-rocmfpx-q8agent-mtp]]:
+`unknown tensor type` ⇒ off the pinned fork; SIGSEGV in
+`ggml_compute_forward_set_rows` ⇒ pre-`36e9acd` build; slow decode with no
+error ⇒ draft not engaged (`draft_n` in `/completion` timings) or an orphaned
+second copy eating GTT (`/sys/class/drm/card1/device/mem_info_gtt_used`);
+500 naming Jinja ⇒ `--jinja` lost; 500 naming the template ⇒ unmapped
+`reasoning_effort` in `models.yml`.

serverAppConfigs-docs/LLM/frank-large-rocmfpx-q8agent-mtp.md

@@ -0,0 +1,232 @@
+---
+type: reference
+title: "frank-large on Dankclaw — ROCmFPX llama.cpp, Qwen3.8-27B Q8_0_ROCMFPX_AGENT + MTP (retired)"
+description: "The retired frank-large runbook — the kingjones30/ROCmFPX llama.cpp fork (pinned commit, gfx1151) serving kingjones777's Q8_0_ROCMFPX_AGENT GGUF, a ftype-115 file no other loader can read, with the repo's own MTP-Q4_0 head, f16 KV, and the BF16 projector. Measured MTP-vs-none A/B, the client contract, and the rollback state as of 2026-10-03."
+tags:
+  - llm
+  - frank-large
+  - dankclaw
+  - llama-cpp
+  - rocmfpx
+  - gfx1151
+  - gguf
+  - spec-decode
+  - mtp
+timestamp: 2026-10-02
+---
+
+# frank-large — ROCmFPX Q8_0_ROCMFPX_AGENT + MTP (gfx1151, retired)
+
+> **RETIRED 2026-10-03** — superseded by
+> [[frank-large-rocmfpx-q6agent-uncensored-mtp]]: Frank swapped to the
+> uncensored Q6_0_ROCMFPX_AGENT file (protected heads, one rung down), which
+> measures **27.6–28.1 t/s** at a deeper prompt than this file's 17.0. This
+> runbook stays live as the rollback reference and the trap catalogue; the
+> payload it documents remains on disk at
+> `/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP/`.
+>
+> Cut over 2026-10-02 from [[frank-large-gufo-qwen38-27b]] (Q4_K_XL, ~15 t/s)
+> on two requests from Frank: *highest quality that runs*, and *one full
+> context*. It delivered 8.39 bpw weights — two quant rungs above what gufo
+> could load — at 17.0 t/s, so quality was not bought with speed.
+>
+> The capability that forced this whole shape is a GGUF tensor type:
+> **`GGML_TYPE_Q8_0_ROCMFPX` (103)**. It exists in exactly one loader on this
+> host, so the runtime was decided by the file, not by preference.
+
+## Identity
+
+| | |
+|---|---|
+| Host / device | Dankclaw (192.168.0.143), Radeon 8060S iGPU `gfx1151`, ~114 GiB GTT aperture |
+| Endpoint | `http://192.168.0.143:7770` — alias `frank-large`, `/v1` OpenAI surface |
+| Runtime | **`kingjones30/ROCmFPX`** llama.cpp fork @ `36e9acd40e10a87cd3c3ef8ec734668757dc8520`, built by [`scripts/build-rocmfpx.sh`](../../scripts/build-rocmfpx.sh) → `/home/dank/GitPackages/kingjones-ROCmFPX/build/bin/llama-server` |
+| Unit | root systemd `frank-large.service` ([`scripts/frank-large.service`](../../scripts/frank-large.service)), `User=dank`, `Restart=always` |
+| Serve script | [`scripts/serve-frank-large.sh`](../../scripts/serve-frank-large.sh) (owns the argv), tunables [`scripts/frank-large.env`](../../scripts/frank-large.env) |
+| Weights | `/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP/Qwen3.8-27B-Q8_0_ROCMFPX_AGENT.gguf` — 28,214,425,952 B, sha256 `f3ef2bdbf56e244f…` |
+| Draft | `…/mtp-Qwen3.8-27B-Q4_0.gguf` — 1,680,271,648 B, sha256 `051a1764cff8c4f3…` |
+| Projector | `…/mmproj-Qwen3.8-27B-BF16.gguf` — 931,145,888 B, sha256 `de2a4986…` |
+| Fetch | [`scripts/download-frank-large-kingjones.sh`](../../scripts/download-frank-large-kingjones.sh) (`-f`, byte-count + LFS sha256 verified, `--verify-only`, token never on the command line) |
+| Seam | `tests/frank_large_unit.bats` (19 tests: payload gates, deployer/serve parity, spec block, KV dtype, sweep identity) |
+
+## Why this file, and why only this runtime
+
+`kingjones777/Qwen3.8-27B-ROCmFP4-STRIX-MTP-GGUF` ships several rungs. Parsed
+headers of the two candidates:
+
+| File | ftype | Tensor mix | Output-head tensors at plain Q8_0 | Card MTP acceptance |
+|---|---|---|---|---|
+| `Qwen3.8-27B-Q8_0_ROCMFPX_AGENT.gguf` | **115** | 353 F32 + 187 Q8_0 + **311 type-103** | **187** | 0.953 |
+| `Qwen3.8-27B-Q8_0_ROCMFPX.gguf` | 111 | 2 Q8_0 + type-103 | 2 | 0.911 |
+
+Two consequences, both measured off the header rather than guessed:
+
+1. **Quality.** The AGENT routing variant keeps the output-head (and other
+   routing-sensitive) tensors at plain Q8_0 instead of folding them into the
+   ROCmFPX kernel layout — 187 vs 2 — so the logits the sampler reads are the
+   highest-precision tensors in the file. It is also the card's MTP-tuned
+   variant, so it is the *faster* drafter base at the same time.
+2. **The runtime is forced.** Type 103 is a private extension: stock llama.cpp
+   and gufo both fail at load (`unknown tensor type`). Verified on the host
+   before deploying — the bake-off was not a horse race, it was a lock-in
+   check.
+
+**The build identity matters as much as the fork.** The 2026-08-25 ROCmFPX
+builds read this file and then died in `ggml_compute_forward_set_rows` — the
+exact failure documented in [[kingjones-q8agent-rocmfpx]] when this payload was
+first attempted on Dankclaw and routed to Dankserver as
+[[frank-medium-rocmfpx]] instead. `scripts/build-rocmfpx.sh` therefore checks
+the commit out **by SHA** and refuses to build anything else; the pin is
+load-bearing, not tidiness.
+
+## Shape and budget
+
+One session, the model's full native window, unquantized KV:
+
+| Dial | Value | Why |
+|---|---|---|
+| `--ctx-size` | 262,144 | "one full context" — the native window, no rope scaling |
+| `--parallel` | 1 | one conversation, all memory to it |
+| KV dtype | **f16** (both K and V) | ~64 KiB/token → ~16 GiB; free inside a 114 GiB aperture, so there is no reason to pay a quant |
+| `--fit off` | on | the auto-fitter would silently resize the window we asked for |
+| `--flash-attn on` | on | required for the long-window attention path to fit |
+| `--direct-io` | on | bypasses page cache for the 26 GiB weight stream (mmap is disabled by it — deliberate) |
+| `--ctx-checkpoints 16` | 16 | slot state for checkpoint/rollback |
+| `--cache-ram 8192` + `--cache-idle-slots` | 8 GiB | prompt cache; verified a repeat pass over the same 6k prompt replays in **0.5 s** |
+| `--jinja` | on | without it `chat_template_kwargs` are silently ignored |
+| `--reasoning on --reasoning-format deepseek --reasoning-budget -1` | | thinking in its own channel, unbounded |
+
+Measured steady state (production, one model copy resident):
+`mem_info_gtt_used` **63.1 GiB** of the ~114 GiB aperture, `mem_info_vram_used`
+0.17 / 2.0 GiB (the iGPU's dedicated carve-out is a stub; everything real lives
+in GTT). llama-server RSS 2.1 GiB. Boot to `/health` 200: **16 s** — the fastest
+endpoint bring-up in the fleet, because there is no engine compile step.
+
+## Speculative decode — measured, not assumed
+
+```
+--spec-type draft-mtp --model-draft <mtp-Qwen3.8-27B-Q4_0.gguf>
+--spec-draft-n-max 4 --spec-draft-n-min 0 --spec-draft-p-min 0.0
+--spec-draft-device ROCm0 --spec-draft-ngl 99
+```
+
+MTP in llama.cpp is a property of the file, not of the flags: the head is a
+trained tensor set (`nextn_predict_layers = 1` in this repo's head, arch
+`qwen35`, 65 blocks) and the draft context runs on the main model's weights, so
+there is no second 26 GiB copy in memory. See the fleet law in
+`AGENTS.md → llama.cpp serving`.
+
+**A/B, 2026-10-02** — same session, temp 0, 2,202-token prompt, 512 tokens,
+`cache_prompt: false`, `frank-large.service` stopped for every scratch arm (the
+benchmarking law), baseline arm stripped of `--model-draft` **and** every
+`--spec-*` flag:
+
+| Arm | Decode t/s (2 runs) | Prefill t/s | Draft acceptance |
+|---|---|---|---|
+| **MTP, n-max 4 (deployed)** | **17.00 / 17.14** | 350.6 / 324.9 | 0.462 (mean acceptance length ≈ 1.86) |
+| no draft | 7.60 / 7.66 | 381.8 / 381.0 | — |
+
+**+123 % decode for an 8–15 % prefill tax.** Notes that matter before someone
+re-litigates this:
+
+- Acceptance here (0.462) is nowhere near the card's 0.953. The card measures
+  its own short, repetitive text; this is a 2.2k-token prose prompt. Treat the
+  card's number as a file property, not a throughput promise.
+- `n-max 4` is where the gain was measured; 2 and 8 are **unmeasured**. The
+  gufo-era law applies in spirit — never rank spec settings on a
+  self-predictable corpus.
+- End-to-end on the chat path is slower than the `/completion` number because
+  thinking tokens are included: a 2,253-token chat prompt generating 512
+  tokens took 43.8 s wall (11.7 t/s end-to-end) — that is the cost of an
+  unbounded reasoning budget, not a regression in decoding.
+
+## Client contract (live-verified 2026-10-02)
+
+| Behavior | Observation |
+|---|---|
+| Model id | `/v1/models` → `frank-large` |
+| Tools | native: `get_weather` → `{"city": "Seattle"}` in `tool_calls` |
+| Thinking | in **`reasoning_content`** (llama.cpp field — *not* vLLM's `reasoning`); `models.yml` sets `reasoningContentField` accordingly |
+| `reasoning_effort` | accepts **only** `xhigh`, `medium`, `low`; anything else is HTTP 500 (the template raises, it does not clamp). `models.yml` folds `high`/`max` → `xhigh` |
+| Sampling | server defaults = the card's thinking recipe (temp 1.0, top_p 0.95, top_k 20, min_p 0, presence 0), mirrored in the harness `extraBody` |
+| Vision | the BF16 projector loads (`loaded multimodal model`); a red square answers **`Red`** (`usage: 75 prompt / 71 cached`) |
+| Prompt cache | second pass over the same 6k prompt: 0.5 s wall |
+
+Three traps, all hit for real:
+
+1. **`content: null` when reasoning eats the budget.** With
+   `--reasoning-parser`-style separation, a request whose `max_tokens` is
+   consumed entirely by thinking returns `message.content = None`. A bench
+   client that indexes `content` crashes *after* measuring and loses the
+   numbers — fall back to `reasoning_content`, or size the budget (200 was
+   enough for the vision probe).
+2. **`MTP boundary missing for seq_id=… ; resyncing after a non-token step`**
+   appears in the log under draft resets. Self-recovering; not an outage signal.
+3. **Never `pkill -f llama-server` on Dankclaw** — frank-small is also a
+   `llama-server`. The sweep key here is the payload name:
+   `pkill -f 'Qwen3.8-27B-Q8_0_ROCMFPX_AGENT'`.
+
+## Operations
+
+```bash
+# deploy (planner selects this endpoint only via its own unit/env/serve trio)
+./scripts/deploy.sh dankclaw --service frank-large
+
+# status / logs / cause of a crash loop
+systemctl status frank-large
+journalctl -u frank-large -n 200 --no-pager
+
+# rebuild the pinned fork (only when deliberately moving the pin)
+scripts/build-rocmfpx.sh
+
+# re-verify the payload without re-downloading
+scripts/download-frank-large-kingjones.sh --verify-only
+```
+
+`serve-frank-large.sh` refuses to start — before touching the port — unless the
+fork binary, the weights, the MTP head and (unless `FRANK_LARGE_MMPROJ=""`) the
+projector all exist; `scripts/deploy-remote.sh` mirrors the same predicate so a
+broken payload never stops the running unit. Caller environment **overrides**
+`frank-large.env`, and defaults use `:=` so an explicitly empty value surfaces
+as the error it reads as.
+
+## Rollback (as of retirement, 2026-10-03)
+
+**Primary rollback: the ALIGNED Q8 file itself is the previous shape's
+payload and remains on disk** at
+`/home/dank/models/kingjones-Qwen3.8-27B-ROCmFP4-STRIX-MTP/` — revert the
+2026-10-03 swap commit (unit/env/serve trio) or repoint
+`FRANK_LARGE_MODEL_FILE`/`LLM_LARGE_MODEL_FILE`, verify with
+`scripts/download-frank-large-kingjones.sh --verify-only`, redeploy.
+
+git tag **`frank-large-gufo-q4kxl-20261001`** (gufo v0.1.1, Q4_K_XL +
+DFlash2) is now a **re-download** rollback: the gufo payloads
+`/home/dank/models/Qwen3.8-27B-UD-Q4_K_XL/` (16.35 GiB) and
+`/home/dank/models/Qwen3.8-27B-UD-Q8_K_XL/` (incl. the DFlash2 drafter) were
+removed from disk on 2026-10-03 with Frank's explicit approval to free space.
+`scripts/build-gufo.sh` remains on disk; re-fetch per
+[[frank-large-gufo-qwen38-27b]].
+
+The one-day vLLM MiMo era's payloads (MiMo-V2.6 BF16 18 GiB,
+Qwen3.5-9B MTP source 19 GiB, both DFlash draft dirs) were staged to
+`/home/dank/.AGENTTRASH/20261002-retired-llm-payloads/` on 2026-10-02 —
+recoverable until the trash is purged, and never part of a rollback plan.
+
+## When it misbehaves
+
+- **`unknown tensor type` / ftype 115 refused** — something moved off the
+  ROCmFPX binary. Check `readlink -f $(systemctl show frank-large -p ExecStart)`
+  and the fork commit; this file cannot run on stock llama.cpp or gufo.
+- **SIGSEGV in `ggml_compute_forward_set_rows` during load** — a fork build
+  older than the `36e9acd` pin; rebuild with `scripts/build-rocmfpx.sh`.
+  History: [[kingjones-q8agent-rocmfpx]].
+- **Slow decode with no log error** — check whether the draft is engaged
+  (`/metrics`, `draft_n` / `draft_n_accepted`; a zero draft count means the head
+  did not load), and check GTT pressure with
+  `cat /sys/class/drm/card1/device/mem_info_gtt_used` — a second resident copy
+  of this model (an orphaned scratch server) drives the whole host into
+  eviction and halves the rate silently.
+- **`500` naming Jinja** — a `chat_template_kwargs` field reached a server
+  without `--jinja`.
+- **HTTP 500 naming the template on the first request of a role** — an
+  unmapped `reasoning_effort`; fix the fold in `models.yml`, not the model.

serverAppConfigs-docs/LLM/frank-large-unsloth-qwen38-27b-q8xl.md

@@ -1,7 +1,7 @@
 ---
 type: reference
 title: "stock llama.cpp Qwen3.8-27B UD-Q8_K_XL (DFlash2) on Dankclaw (frank-large)"
-description: "The active frank-large runbook — the stock llama.cpp prebuilt (unsloth b11030, ROCm gfx1151) serving unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL (27B dense, 31.46 GiB) with the z-lab DFlash2 Q4_K_M drafter (depth 4) and q8_0 KV on the Dankclaw Strix Halo iGPU: the prebuilt runtime, the full-window/one-stream shape, the DFlash2 spec method, the Qwen-Sharp template, the payload, the client contract, and the failure modes worth knowing."
+description: "The RETIRED frank-large runbook (2026-09-21 → 2026-09-28; replaced by [[frank-large-gufo-qwen38-27b]], and the rollback target) — the stock llama.cpp prebuilt (unsloth b11030, ROCm gfx1151) serving unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL (27B dense, 31.46 GiB) with the z-lab DFlash2 Q4_K_M drafter (depth 4) and q8_0 KV on the Dankclaw Strix Halo iGPU: the prebuilt runtime, the full-window/one-stream shape, the DFlash2 spec method, the Qwen-Sharp template, the payload, the client contract, and the failure modes worth knowing."
 tags:
   - llm
   - llama.cpp
@@ -19,10 +19,12 @@ timestamp: 2026-09-21
 
 # frank-large — stock llama.cpp Qwen3.8-27B UD-Q8_K_XL (DFlash2)
 
-> **ACTIVE since 2026-09-21.** Replaces [[frank-large-kairic-edge-qwen38-27b]]
-> (Kairic Edge IU4 W4A4, now the rollback target) and
-> [[frank-large-unsloth-qwen38-flash-next]] (the Flash-Next MoE it preceded).
-> Same port (7770), same alias (`frank-large`), same client contract surface.
+> **RETIRED 2026-09-28**, replaced by [[frank-large-gufo-qwen38-27b]] (the
+> gufo engine, same weights, same port and alias); the prebuilt is intact on
+> disk as the rollback target. ACTIVE from 2026-09-21, when it replaced
+> [[frank-large-kairic-edge-qwen38-27b]] (Kairic Edge IU4 W4A4) and
+> [[frank-large-unsloth-qwen38-flash-next]] (the Flash-Next MoE it
+> preceded).
 
 ## Identity
 

serverAppConfigs-docs/LLM/frank-large-vllm-mimo-v26-bf16.md

@@ -0,0 +1,186 @@
+---
+type: reference
+title: "frank-large on Dankclaw — vLLM TheRock, MiMo-V2.6-Distill-Qwen-9B BF16 + cross-model MTP (RETIRED)"
+description: "The RETIRED frank-large runbook (2026-10-01, one day) — the digest-pinned kyuz0 TheRock gfx1151 vLLM image serving XiaomiMiMo MiMo-V2.6-Distill-Qwen-9B in full BF16 on the Strix Halo iGPU. Kept for the measured record: the roof math, the cross-model MTP drafter, and the autoresearch finding that spec decode is NET-NEGATIVE on real chat+reasoning traffic here (6.19 t/s with MTP-2 vs 9.01 t/s without). Rolled back to [[frank-large-gufo-qwen38-27b]] by Frank the same night."
+tags:
+  - llm
+  - frank-large
+  - dankclaw
+  - vllm
+  - strix-halo
+  - mimo
+  - mtp
+  - spec-decode
+  - bf16
+timestamp: 2026-10-01
+---
+
+# frank-large — vLLM MiMo-V2.6-Distill-Qwen-9B BF16 (cross-model MTP)
+
+> **RETIRED 2026-10-01 night, one day active.** Frank's call after the
+> autoresearch loop: battery decode 6.19 t/s (MTP-2) / 9.01 t/s (no draft)
+> against gufo's ~15 t/s on the 27B — and every ≥6-bit quant path in vLLM
+> 0.30 is dead on gfx1151 (quark int8 rejected; all FP8 kernels CDNA-only;
+> torchao config unwired). Back to [[frank-large-gufo-qwen38-27b]]
+> (Q4_K_XL, 1 x 262144). The measured content below stands as the record.
+
+## Identity
+
+- **Endpoint:** `frank-large` at `192.168.0.143:7770` (Dankclaw).
+- **Runtime:** vLLM 0.30.0 in the **digest-pinned** `docker.io/kyuz0/vllm-therock-gfx1151`
+  image (image ID `2d24a8c5a3fd`, digest `sha256:98c1f3ac9833…` — ROCm 10.0.0 /
+  torch 2.11.0 / TheRock, Triton Gated DeltaNet kernels). The official vLLM
+  images are CUDA-only; this fork is the fleet's gfx1151 path. Plain
+  `podman run` driven by `scripts/serve-frank-large.sh` under a root unit
+  (`scripts/frank-large.service`, `User=dank`, `Restart=always`,
+  `ExecStop=podman stop -t 120`, `TimeoutStopSec=180` > the stop grace).
+- **Model:** `XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B` in **full BF16**
+  (17.53 GiB, 4 shards) — the same checkpoint frank-small serves as an INT8
+  W8A16 AutoRound conversion on the 3090, here unquantized.
+- **Draft:** `Qwen/Qwen3.5-9B` (18 GiB on disk; only the 15 `mtp.*` tensors
+  are loaded into the draft pass).
+
+## Why this shape (the roof math)
+
+Decode speed for a dense model on Strix Halo is bandwidth-bound:
+BF16 weights x one pass per token against the LPDDR5x aperture. Measured
+2026-10-01 on this host: **13.1 t/s is ~94% of the roof** — the payload, not
+a misconfiguration. The same INT8 payload reaches 83.9 t/s on the 3090
+because it moves ~12 GiB through ~936 GB/s. The only way past a bandwidth
+wall is generating more than one token per weight pass: **speculative decode
+is the performance feature of this endpoint**, not a garnish.
+
+The memory side is the opposite of a squeeze: KV grows ONLY on the 8 full
+attention layers (32 KiB/token bf16: 8 x 2 x 4 heads x 256 dim x 2 B);
+the 24 Gated DeltaNet layers hold a FIXED ~25 MiB state per running
+sequence. `--max-model-len 262144` (the native window) with
+`--max-num-seqs 4` at `--gpu-memory-utilization 0.45` boots clean — the
+four-session contract the gufo shape ran, with the pool shared by the
+scheduler instead of carved per session.
+
+## Spec method: cross-model MTP — measured, not assumed
+
+| arm | prefill | decode | acceptance |
+|---|---|---|---|
+| BF16 base | 17,876 t/s | 13.1 t/s | — |
+| BF16 + DFlash 8 tok | 7,976 t/s | 17.4 t/s | pos0 59% pos1 26% (token rate 14.8%) |
+| **BF16 + MTP 2 tok** | 7,380 t/s | **19.8 t/s** | pos0 71% pos1 44% (token rate 57.4%) |
+
+One session, temperature 0, 28,647-token prefill / 512-token decode, same
+client for all three arms (2026-10-01 bench window, frank-large stopped per
+Benchmarking Policy). **MTP wins (+51% vs +33%) and needs only 2 draft
+tokens** — acceptance past position 1 is 44% and falling; DFlash's own curve
+decays 59/26/14/7, so 8 proposals are mostly wasted verification.
+`num_speculative_tokens` beyond 2 is UNMEASURED; do not raise it without a
+fresh A/B. Both spec arms pay ~55–59% of prefill (the draft passes share
+the compute queues) — this endpoint's thinking-heavy traffic spends its wall
+clock in decode, which is the regime where that trade wins.
+
+**Why a Qwen head accepts on a MiMo target:** MiMo-V2.6 is a distillation of
+Qwen3.5-9B that preserves the hidden-state geometry — identical
+4096-hidden/248320-vocab/32-layer/interval-4 structure, and the 71%/44%
+per-position acceptance is the direct measurement that the distillation
+preserved more than geometry. MiMo itself ships **ZERO `mtp.*` tensors**
+(verified in both the BF16 base and the INT8 conversion) — which is exactly
+why frank-small's SGLang NEXTN draft is a 0.00-acceptance ghost: SGLang
+reads the head from the TARGET checkpoint, vLLM reads the draft from a
+separate path. That difference is the whole trick:
+
+```
+--speculative-config '{"method":"mtp","model":"/models/Qwen3.5-9B","num_speculative_tokens":2}'
+```
+
+The alternative drafter (`z-lab/Qwen3.5-9B-DFlash`, 6 layers, target_layer_ids
+[1,5,…,29]) stays fetched on disk; flip `FRANK_LARGE_SPEC=dflash` and repoint
+`FRANK_LARGE_SPEC_DIR` to use it. Its card's "vLLM PR #40898 unmerged" claim
+is FALSE — `vllm/v1/spec_decode/dflash.py` ships in 0.30.
+
+## The numerics trap (greedy text is NOT preserved by spec decode)
+
+Measured: temperature-0 completions DIFFER across arms (sha `1318aeb6` base,
+`0208b45d` dflash, `20c41c38` mtp). Batched verification changes the bf16
+reduction order, argmax near-ties diverge, and the divergence cascades.
+vLLM spec decode is distribution-preserving, not bitwise-greedy-preserving.
+The gufo-era "greedy == autoregressive text" law does not carry over; do not
+chase text drift between spec and no-spec arms as a regression.
+
+## Payload
+
+- Target: 4 safetensors shards (17.53 GiB) + config/tokenizer under
+  `/home/dank/models/MiMo-V2.6-Distill-Qwen-9B-BF16/`, each shard
+  byte-count + LFS sha256-verified by
+  `old/download-frank-large-mimo.sh` (`--verify-only` re-checks; no token on
+  the curl command line).
+- MTP source: `old/download-frank-large-qwen35.sh` →
+  `/home/dank/models/Qwen3.5-9B/` (4 shards behind
+  `model.safetensors.index.json`; the shards carry the odd HF asset names
+  `model.safetensors-0000N-of-00004.safetensors` — the index resolves them,
+  the engine boots from exactly this tree).
+- DFlash drafter: `old/download-frank-large-dflash.sh` →
+  `/home/dank/models/Qwen3.5-9B-DFlash/`.
+- **Payload status 2026-10-02:** all three trees moved to
+  `/home/dank/.AGENTTRASH/20261002-retired-llm-payloads/` (59 GiB staged with
+  the other retired frank payloads) and the fetch trio moved to `old/` — the
+  endpoint is two runtimes past this one, and re-fetch is cheaper than disk.
+- The serve script gates the image by REPO DIGEST (rootless podman resolves
+  tags from the network; the pin exists to boot exactly these bytes
+  offline), then config + index + exact shard count, then the spec draft
+  (either weight layout: single `model.safetensors` or a sharded index),
+  then `/dev/kfd` + whole `/dev/dri` (never individual renderD nodes —
+  Dankserver 2026-09-29 lesson).
+
+## Client contract
+
+- **Alias:** `frank-large` (the `id` at `/v1/models`).
+- **Reasoning:** `--reasoning-parser qwen3`; vLLM emits the thinking text in
+  **`reasoning`**, NOT llama.cpp/gufo's `reasoning_content` — the harness
+  `models.yml` compat block for this provider sets
+  `reasoningContentField: "reasoning"`. Miss that and thinking silently
+  vanishes from transcripts.
+- **Tools:** `--enable-auto-tool-choice --tool-call-parser qwen3_coder`.
+- **Vision:** the checkpoint carries the `qwen3_5_vision` tower but the
+  serve script passes `--language-model-only` — the endpoint stays
+  text-only by decision; dropping the flag brings vision back.
+- **Sampling:** request-side; the harness sends the Qwen3.5-family recipe
+  (temp 0.7, top_p 0.8) via `compat.extraBody`, and
+  `chat_template_kwargs.{enable_thinking,preserve_thinking}` — the
+  `chat_template_kwargs` form is what vLLM reads; a top-level
+  `preserve_thinking` is ignored there.
+- **Env hygiene:** every tunable is `FRANK_LARGE_`-prefixed because HIP
+  reads bare `TEMP`/`HOST`/`GPU_MAX_HEAP_SIZE` and the serve script exports
+  the env file with `set -a`.
+
+## Ops
+
+- Boot ~200–240 s warm; health 503 while loading (do not kill it).
+- The serve script sweeps stale container identities before starting:
+  `frank-large-mimo` itself and the bench-window container `vllm-mimo`.
+- Container death → `podman wait` returns non-zero from ExecStart →
+  `Restart=always` self-heals (a clean exit-0 drain also restarts, the
+  frank-primary lesson).
+- Bench windows: `systemctl stop frank-large` is enough (mask fails on the
+  relinked unit; the deploy restores).
+- First-deploy facts (2026-10-01, deploy #3): healthy after **230 s**;
+  boot log `GPU KV cache size: 819,332 tokens, Maximum concurrency for
+  262,144 tokens per request: 3.13x` — four sessions share the pool
+  (the scheduler, not a carve). Live battery green: alias `frank-large`,
+  reasoning split into `reasoning`, tool_calls with correct args, and
+  `/metrics` acceptance pos0 69% pos1 44% (54/86 draft tokens accepted),
+  matching the bench window.
+
+## Rollback
+
+Git tag **`frank-large-gufo-q4kxl-20261001`** — gufo v0.1.1 (built from
+source, `scripts/build-gufo.sh` stays on disk) + the Q4_K_XL Qwen3.8-27B
+payload + DFlash2 drafter, all still on the host. Restoring the tag's
+unit/env/serve trio and redeploying returns the 27B endpoint.
+
+# Citations
+
+- Model: https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B
+- Draft source: https://huggingface.co/Qwen/Qwen3.5-9B
+- DFlash drafter: https://huggingface.co/z-lab/Qwen3.5-9B-DFlash
+- Engine image: https://hub.docker.com/r/kyuz0/vllm-therock-gfx1151
+- Serve script: `scripts/serve-frank-large.sh`; tunables: `scripts/frank-large.env`
+- Predecessor: [[frank-large-gufo-qwen38-27b]]; sibling on the same arch:
+  [[frank-small-sglang-mimo-v26]]

serverAppConfigs-docs/LLM/frank-medium-batch-ubatch.md

@@ -134,6 +134,59 @@ slots of 131072, `6.44 / 6.90 GiB` free — matching the pre-deploy measurement
 0.01 GiB — a tool call and two concurrent requests answered, draft acceptance on
 both slots (0.63 / 0.48 on slot 1, 0.39 on slot 0), and no ROCm errors.
 
+## The split: re-measure it at the deployed shape
+
+The batch and the tensor split are **coupled**: compute buffers are not
+symmetric across the two cards, so any batch change moves the balance point.
+Order of work — sweep batch with llama-bench (above), THEN re-measure the
+split at the DEPLOYED shape (262144, two slots, drafter loaded) by loading
+the real service with a per-config env override and reading `rocm-smi`. A
+llama-bench number cannot decide the split: it has no `--parallel`, sizes
+its own context as `n_prompt + n_gen + n_depth`, and cannot load the drafter.
+
+The settle gate is mandatory (above); wait for BOTH cards under ~1 GiB AND
+the port to be free (`ss -ltn | grep :7770` until gone) before the next
+load, or the transient server dies with `couldn't bind HTTP server socket`.
+Free bytes follow `card = weights_share + KV_share + constant + ubatch_term`,
+where the ubatch term costs the drafter-carrying card roughly **1.85×** the
+other (measured 2.41 GiB vs 1.30 GiB going 1536 → 512). Balance by
+measurement only: one `-ts` step is one layer, worth ~0.43 GiB on a DeltaNet
+boundary and ~1.43 GiB on an attention one, so the reachable minimum gap is
+about one DeltaNet layer, never zero. Reference: 36,29 @1536 → 4.71/4.91;
+36,29 @512 → 6.01/7.32; **35,30 @512 → 6.44/6.90 (deployed)**.
+
+Re-fit both ubatch coefficients in `tests/frank_medium_unit.bats` at the new
+deployed point and re-anchor the per-card constants; check neighbouring
+splits still FAIL the balance gate or the seam has stopped discriminating
+(keep the 1.5 GiB/card floor and the one-DeltaNet-layer slack). Deploy,
+then verify live: flags in `/proc/<pid>/cmdline`, `/slots`, VRAM matching
+the pre-deploy measurement, a tool call, two concurrent requests,
+`draft acceptance` lines on BOTH slots, 0 ROCm errors.
+
+**Per-card utilization (who is hot, and the split lever)** — a different
+question from batch/prefill. Measure decode tok/s AND per-card
+`rocm-smi --showuse` at 2–3 neighbouring splits in ONE session. Pitfalls
+(each cost a re-run): run the transient unit as ROOT (a `pkill -9
+llama-server` as dank auto-restarts the service and `systemctl` as dank
+fails "Access denied"); warm up 3 throwaway requests before the timed one
+(first 384-token decode after load measured 1.77 tok/s against a true ~32);
+`rocm-smi --showmeminfo vram --csv` columns are `$2`=Total, `$3`=**Used** —
+parsing the wrong column spins the settle gate the full 600 s; the util% is
+sampled post-hoc (which card finished last), so treat it as "which card is
+the pipeline tail", not a duty cycle.
+
+**What the split does and does not buy (measured 2026-09-17, 393216 /
+MTP n=3):** the split is NOT a decode-throughput lever — 34:31, 35:30, 36:29
+all sit in the same ~32 tok/s band. The x1 card runs hot (~91–97 %) at
+35:30 because the MTP drafter + output layers + heavier compute buffers are
+pinned to the card holding the LAST layers — not because PCIe is the
+bottleneck (~5.2 MB of boundary activation crosses per 512-token ubatch).
+36:29 makes the x16 card the utilized worker (42 %/3 %) at no throughput
+cost but spends headroom (0.45 GiB free, under the floor) — Frank's standing
+preference is that the card with the FASTEST PCIe lane should be the more
+utilized one; deploying it requires re-fitting the seam's slack/floor and
+re-checking that 35:30 and 37:28 still fail so the gate keeps discriminating.
+
 ## Limitations
 
 - **`llama-bench` cannot run the drafter, the slots or the deployed context.**

serverAppConfigs-docs/LLM/frank-primary-freetoken-pr587.md

@@ -0,0 +1,170 @@
+# frank-primary — FreeToken PR #587 ROCm (Qwen3.8-Flash-Next-Uncensored-FP8, 1× R9700)
+
+**Status: ACTIVE 2026-10-05 →** · **Host:** Dankserver (192.168.0.235) · **Port:** 7770 · **GPU:** 1× Radeon AI PRO R9700 (gfx1201) — the second R9700 sits idle
+
+The tcclaviger RDNA4 vLLM fork was replaced by **FreeToken at PR #587 head
+`77a7201f98f00d93f618c6cf8b81ffe8360e0899`** (branch `rocm714_torch214`), built
+on the host into the local image
+`localhost/frank-primary-freetoken:pr587-77a7201` by
+`scripts/docker/frank-primary-freetoken.Dockerfile`. Same port, same served
+alias (`frank-primary`), same 131-shard payload
+(`/home/dank/models/Qwen3.8-Flash-Next-Uncensored-FP8`). Rollback:
+git tag `frank-primary-vllm-shapeG-20261003` (the vLLM trio + its env lore),
+runbook [[dankserver-qwen38-fp8-expert-offload-tuning]].
+
+## Why FreeToken
+
+The vLLM fork planned a per-boot VRAM/host tiering from an estimator: util
+0.982 was a hard law (1.00 was **unbootable** — worker init raises
+`ValueError: Free memory on device cuda:N (31.29/31.86 GiB)` when budget ==
+total), the expert arena capped at 17.77 GiB / 7,762 slots, and live traffic
+still degraded ~40k tokens in (the GDN state-fault class, never fixed in the
+fork). FreeToken keeps **all expert weights in pinned host RAM** with an LRU
+GPU cache, sizes **KV as an explicit token pool**, and boots in one
+deterministic pass — no per-boot estimator to fight.
+
+## The build (on dankserver, rootless podman as dank)
+
+```bash
+cd /home/dank/serverAppConfigs
+podman build -t frank-primary-freetoken:pr587-77a7201 \
+  -f scripts/docker/frank-primary-freetoken.Dockerfile scripts/docker/
+```
+
+Laws baked into the Dockerfile (each one cost a failure):
+
+- **Base digest-pinned**: `rocm/pytorch:rocm7.14_ubuntu24.04_py3.12_pytorch_release_2.11.0@sha256:a223aee1…` —
+  the PR doc's known-good stack (torch 2.11.0+rocm7.14, triton 3.7). No
+  rocm7.14+torch2.14 image exists on Docker Hub (the 2.14 images are
+  rocm10.1.0 — an untested pair for this PR).
+- **`--no-build-isolation --no-deps -e .`** — a plain
+  `pip install -e .` on ROCm can REPLACE the image's ROCm torch with a CUDA
+  build (the dependency set is authored for NVIDIA). Runtime deps are
+  curated explicitly; `flashlib==0.3.0 --no-deps` (its numba /
+  nvidia-cutlass-dsl deps are CUDA-only); ninja for the QSA JIT;
+  **flashinfer / sgl_kernel intentionally absent** (no ROCm builds;
+  FreeToken falls back to its Triton kernels — the PR's 12 pytest failures
+  are exactly those backends).
+- **`libhiprtc.so` dev symlink**: the pip-ROCm torch ships only the
+  versioned `libhiprtc.so.7` in
+  `/opt/venv/.../site-packages/_rocm_sdk_core/lib`, but the PR's `_cpu_moe`
+  extension links `-lhiprtc`. First build died `ld: cannot find -lhiprtc`;
+  the Dockerfile symlinks the unversioned name into the ALREADY-LINKED dir.
+- FreeToken source is checked out at the **pinned PR head**, not the branch
+  (a future push to `rocm714_torch214` cannot change the deployed engine).
+- The stale Docker Hub login (`docker login` creds at
+  `~/.docker/config.json`) poisons podman's ANONYMOUS pulls of public images
+  with "invalid username/password" — `docker logout` fixes the pull. Hit
+  2026-10-05 while pulling the base image.
+
+## Measured at cutover (2026-10-05/06)
+
+
+- **Boot**: ~2.3 min container-start → API-ready (131 shards page-in + PLE
+  pin + expert staging + CUDA graphs `[1,2]`). VRAM after graphs:
+  **1.80 GiB free** of 31.86. RAM: 122 GB used + **113 GB shared (the
+  pinned expert/PLE slabs)**, 123 GB still available.
+- **Decode: 17.1 tok/s** single-stream, temp 0, 400-token probe (clean
+  1→N counting). The vLLM incumbent measured ~30 tok/s on TWO cards;
+  FreeToken single-card is the deployed baseline.
+- **The single-card LRU ceiling**: the KV contract (harness worst case
+  177,356 tokens) + dense fp8 + the vision tower + graphs leave ~12 GiB for
+  the expert LRU at rate 0.10 — and only 1.8 GiB of live headroom, so
+  `ft ctl cache --moe` cannot raise it far on one card. The paths to a
+  better decode rate are TP=2 (below) or a smaller pool (contract
+  violation — the 177,356 floor is real traffic, not padding).
+- **TP=2 exists but does not boot on ROCm** (probed 2026-10-06,
+  zero-impact window): `--tensor-parallel-size 2 --gpu 0,1` dies in the TP
+  allreduce JIT — `pynccl.cu` compiles per-rank at boot and clang can't
+  resolve `cuda_runtime.h` from the hipified include path
+  (`1 error generated when compiling for gfx1201`, ninja stopped). The PR
+  author validated single-GPU only; this is FreeToken's untested-TP-on-ROCm
+  edge, not a board or driver fault (radiance TP=2 allreduces fine on this
+  pair since the 2026-09-20 swap). A future fix = patch the JIT include
+  path (HIP's cuda_runtime shim) or vendor an ROCm allreduce — re-probe
+  then.
+- **Verification gates passed**: /health ok; plain completion `FLEET-OK`
+  (finish stop); thinking split — `reasoning_content` non-empty AND
+  `content` carries the answer (17×24 → "408"); native `tool_calls` with
+  correct name+arguments (finish tool_calls); 256×256 solid-red PNG through
+  an `image_url` part answers exactly "Red" (prompt_tokens 127); stats:
+  native ctx 262,144, KV 4,687 pages × 64 = 299,968, modalities
+  [text,image], sampling defaults from generation_config.json
+  (temp 1.0 / top_p 0.95 / top_k 20).
+- **Thinking counts against max_tokens** (same as the vLLM era): a
+  `max_tokens: 10` probe returns EMPTY content with the budget spent in the
+  think block. Clients must budget for thinking or disable it.
+- **Prometheus scrapes /metrics → 404** (Dankclaw scrapes 192.168.0.235:7770
+  from the llm-fleet job; FreeToken has no /metrics endpoint). The target
+  will read down until the scrape config is repointed or a metrics shim
+  lands — same class as the retired HTML-probe targets
+  (see [[Observability]]).
+
+
+## The shape
+
+| Knob | Value | Why |
+|---|---|---|
+| `--gpu` / `HIP_VISIBLE_DEVICES` | 0 / 0 | ONE R9700; visibility pinned so in-container index 0 is the right card |
+| `--num-tokens` | **299,968** | KV pool in TOKENS; **must be a multiple of 64** (the auto attention backend is `qsa_sparse`, which overrides page size to 64 — 300,000 was refused: `nearest valid values: 299968 or 300032`) |
+| `--max-running-requests` | 2 | first-boot conservative; raise after real-traffic slot-occupancy look |
+| `--max-output-tokens` | 36,864 | harness maxTokens; requests that omit max_tokens get this default |
+| `--moe-cache-rate` | 0.10 | GPU expert LRU as fraction of all experts (~120 GiB fp8 → ~12 GiB VRAM); first-boot conservative |
+| parsers | AUTO | thinking → `reasoning_content` natively; tools in the qwen3.8 dialect; NO `--reasoning-parser`/`--tool-call-parser` flags |
+
+KV math: 12 full-attention layers × 2 KV heads × 256 head_dim × 2 (K+V) ×
+2 B = **24 KiB/token** (f16 KV; the 36 GDN layers use fixed per-seq state,
+not per-token KV) → 299,968 × 24 KiB ≈ 7.0 GiB. Clears the harness worst
+case (140,492-token prompt + 36,864 max_tokens = 177,356) and admits ONE
+native 262,144-token conversation + full budget. `models.yml`
+`contextWindow: 200704` stays (compacts early, never 400s).
+
+## RAM math (why the root unit + memlock)
+
+246 GiB total. FreeToken pins ~120 GiB of fp8 expert rows + the ~47.7 GiB
+PLE ngram table in host RAM → the unit carries
+`LimitMEMLOCK=infinity` (root unit; dank's login session caps at 8 MiB and
+a user quadlet cannot raise it). Weights+PLE+overhead ≈ 190 GiB — the vLLM
+container's ~170 GiB MUST be down first; `systemctl stop frank-primary`
+frees it (measured: 207 GB free right after the stop).
+
+## Resize live
+
+```bash
+ft ctl cache --kv <tokens>   # grows the KV pool LIVE (shrink needs restart)
+ft ctl cache --moe <rate>    # same for the expert LRU
+ft ctl stats                 # pool occupancy, VRAM, throughput
+```
+
+## Verification (the fleet's real-completion law)
+
+1. `/health` 200 — necessary, not sufficient.
+2. Plain completion returns the asked-for string.
+3. Thinking request: `reasoning_content` non-empty AND `content` carries the
+   answer (the models.yml `reasoningContentField: reasoning_content` flip
+   rides on this).
+4. Vision probe: a solid-color PNG through an `image_url` part answers the
+   exact color (the legacy top-level `images` field trap does not exist
+   here — FreeToken is OpenAI-native).
+5. Unit-managed: `systemctl show -p MainPID` behind the health answer (the
+   2026-09-26 stray-port false green).
+
+## Rollback
+
+```bash
+git checkout frank-primary-vllm-shapeG-20261003 -- \
+  scripts/frank-primary.env scripts/serve-frank-primary.sh \
+  scripts/frank-primary.service
+./tests/run.sh && git commit -m "frank-primary: rollback to vLLM shape G" \
+  && git push && ./scripts/deploy.sh dankserver --service frank-primary
+```
+
+The tcclaviger image and the FreeToken image coexist in dank's rootless
+storage; the deploy re-pulls the pinned digest if absent.
+
+## Index
+
+- [[dankserver-qwen38-fp8-expert-offload-tuning]] — the retired vLLM shape
+  (rollback target; its 362-line env lore lives in the tag)
+- [[freetoken-frank-small]] — the v0.1.2 CUDA precedent (ortobox, RTX 3090)
+- [[LLM/index]] — fleet index

serverAppConfigs-docs/LLM/frank-primary-glm53-flash-w4a16.md

@@ -0,0 +1,798 @@
+---
+type: runbook
+title: frank-primary — GLM-5.3-Flash W4A16 int4 (tcclaviger, 2x R9700)
+description: VERIFIED 2026-09-30 by boot probe, bench, live expert-cache telemetry AND a torch-profiler kernel-time window. frank-primary cut over from Qwen3.8-Flash-Next-Uncensored-FP8 + MTP4 to Intel/GLM-5.3-Flash-W4A16-AutoRound, the first payload this fleet serves through the fork's W4A16 expert-offload path (INC scheme -> R4dW4a16MoEMethod, libr4d grouped W4A16 GEMM). Measured decode 9.9-10.0 tok/s and prefill 918 tok/s at 131,072 ctx. THE 35 tok/s TARGET IS NOT MET, and the profiler window FALSIFIED the first explanation: the expert gather is 1.05 % of device time, not the bound. 44 % of the 93 ms/token budget is pointwise glue (1,428 elementwise launches per token) because the fork's GLM trunk is NOT torch.compiled while its own Qwen trunk is; the second-largest lever, the fused router, is binary-blocked at 288 experts. The compile changelist is staged but incomplete (three patches, one blocker left).
+tags:
+  - dankserver
+  - r9700
+  - gfx1201
+  - glm5.3
+  - w4a16
+  - int4
+  - expert-offload
+  - expert-cache
+  - autoround
+  - vllm
+timestamp: 2026-09-30
+---
+
+# frank-primary — GLM-5.3-Flash W4A16 int4 (tcclaviger, 2x R9700)
+
+> **RETIRED 2026-10-01.** Cutover reversed to Qwen3.8-Flash-Next-Uncensored-FP8
+> + MTP4 (commit `2339fae` trio). The DFlash2 drafter experiment is a
+> **failure**: after five patches the drafter served and drafted (12.8 tok/s
+> proposed, 65,536 ctx, both TP ranks capturing FULL dflash2 cuda graphs) but
+> **acceptance pinned at 1.00-1.38 on every text class including verbatim
+> repetition** — the signature of a mis-wired aux-hidden-state capture, not a
+> weak drafter. At ~1.2 acceptance the 8-token verification tax makes decode
+> 5x SLOWER than the 9.9-10.0 no-draft baseline (1.6-1.8 tok/s measured). The
+> untested alternative capture semantics: post-layer output state
+> (DeepSeek's `hidden_states + residual`) or the mHC `comb` stream. Payload
+> (170 GiB) + drafter (2.2 GiB) staged to
+> `/home/dank/.AGENTTRASH/20261001-glm-purge/` pending purge approval.
+> KV-cache work that survives: patches 02/03/04 in `scripts/patches/` are
+> generic unifier/tailpin/grouping fixes, apply-verified against the image.
+
+
+> **Question this page answers.** Can the fleet serve the 320B-class
+> `GLM-5.3-Flash` at int4 on two R9700s with expert offload, and how fast is it?
+> **Answer.** Yes — it serves, with vision, tool calls and 131,072 ctx. It runs
+> at **9.9-10.0 tok/s decode / 918 tok/s prefill**, and **35 tok/s is not
+> reachable** on this payload as the fork ships it. The bound is **not** the
+> model, the quant, the link, or the expert gather (measured: 1.05 % of device
+> time). It is the **unfused pointwise glue of an uncompiled trunk** — 44 % of a
+> 93 ms per-token budget spread over ~1,428 elementwise launches per token —
+> because the fork compiles its own Qwen trunk and not its GLM one. See
+> [The measured profile](#the-measured-profile-and-the-real-wall).
+
+## DFlash2 target-side port — the four laws and the KV ladder
+
+Distilled 2026-10-05 from the probe windows (skills `glm53-dflash-boot-ladder`
+/ `glm53-dflash-eagle3-port`); each law was proven by a real failed boot.
+
+**The four EAGLE3-port laws** (make the GLM trunk satisfy `SupportsEagle3`):
+
+1. **Patch the instance the runner calls.** `model_runner.py:425` calls
+   `set_eagle3_aux_hidden_state_layers(self.model)` on the *loaded* model —
+   GLM loads as `Glm5NextForConditionalGeneration` (multimodal wrapper;
+   `self.language_model = Glm5NextForCausalLM`). Members placed only on the
+   CausalLM fail the check on the conditional — same error text, wrong file.
+   Patch BOTH wrappers.
+2. **Presence-checked member set.** `SupportsEagle3` (interfaces.py:1659) is
+   `@runtime_checkable`; `isinstance` walks ALL `__protocol_attrs__`
+   (including base-protocol members) with `getattr_static`, breaking on the
+   first miss. Both wrappers need, as class attributes:
+   `supports_eagle3: ClassVar[Literal[True]] = True`,
+   `has_own_lm_head: bool = False`, `has_own_embed_tokens: bool = False`.
+   Verify in-image with stock typing — no boot needed.
+3. **MRO forbids inheriting the Protocol.** GLM wrappers carry four/five
+   Protocol-derived bases; adding `SupportsEagle3` anywhere dies with
+   "Cannot create a consistent method resolution order" (sim-proven: the
+   ordering rejects it in every base position). The attribute route avoids
+   the MRO, so the two methods must be written by hand: the CausalLM
+   unwraps to `self.model` (the `EagleModelMixin` trunk), the conditional
+   delegates to `self.language_model` (pixtral.py:498 pattern);
+   `get_eagle3_default_aux_hidden_state_layers` returns `(2, L//2, L-3)`,
+   with the drafter config's `target_layer_ids` winning when present.
+4. **Trunk capture point:** append `aux_hidden_states` BEFORE the layer call
+   — GLM's loop carries the single-stream state entering the layer.
+
+**The KV blocker ladder** (patches `scripts/patches/glm53-dflash-0*`):
+
+1. EAGLE3 gate → presence-checked attrs on both wrappers (above).
+2. KV page-size unifier: MLA layers cannot be padded (sparse MLA reads
+   whole token rows) but CAN grow block_size when the max page is an exact
+   multiple — bump `max_page_size` in `unify_kv_cache_spec_page_size` to
+   the `lcm` of all MLA-family pages first (patch 02).
+3. kpool tail assert `tail_kv_cache.shape[2] == pool_size`: the unifier's
+   divisible branch grew KpoolTailSpec's block; its kernel pins block ==
+   index_kpool. Add a KpoolTailSpec branch that pads `page_size_padded`
+   instead of growing (patch 03).
+4. KV capacity: the glm5 layout ALIASES mamba states INTO MLA pages (why
+   GLM fits 32 GiB cards); the generic path (taken when draft/hidden specs
+   exist) cannot alias → ~30 GiB demanded vs a 0.92 GiB reserve. Extend
+   `_get_kv_cache_groups_glm5_next` to partition
+   HiddenStateCacheSpec + drafter attention specs; the drafter forms its
+   own uniform group (per-group pages, NO unification → the MLA block
+   stays at the engine-auto-set 2,432 from interface.py:918, kpool guard
+   `block_size % 512 == 0` unstressed); layout returns a 10-tuple
+   (+draft_group, +hidden_groups); the builder sizes ONE shared backing
+   allocation (`allocate_kv_cache` asserts all KVCacheTensor sizes equal) =
+   aliased pool + appended draft/hidden tensors (patch 04). After 04 the
+   capacity requirement collapsed 31.76 → 1.88 GiB at 65,536 max len and
+   the engine SERVED (READY ~500 s, dflash2 fused prep ON, drafter FULL
+   graphs on both TP ranks).
+5. Reserve registry: `_glm53_kv_bytes_per_token` must add 4,096 (drafter
+   KV) + 5×hidden×2 (aux hidden cache) keyed on
+   `speculative_config.method in {dflash, dspark, extract_hidden_states}`.
+   NEVER key on `use_aux_hidden_state_outputs` — it lives on the
+   gpu_model_runner instance (gpu_model_runner.py:623/678), unset at plan
+   time, so the check silently fails and the reserve stays 13,600
+   (patch 04b). Proven reserve math at 65,536: 13,600 + 4,096 +
+   5×4096×2 ≈ 58,656 B/token → ~4.15 GiB reserve, arena ~10.3 GiB.
+
+**Verify cheaply before booting:** smoke-test the pure KV logic on CPU
+torch with the image's real spec classes — stub `vllm.logger`/`vllm.utils.*`,
+register the spec classes in `KVCacheSpecRegistry` with a dummy manager, add
+`from __future__ import annotations`, and run synthetic GLM+draft+hidden
+spec sets through the patched grouping/layout/builder functions. A 7-minute
+boot per hypothesis is the alternative.
+
+**Window discipline:** stop + mask `frank-primary` FIRST and gate on
+MemAvailable ≥ 230 GiB before booting the probe (the expert-offload planner
+reads MemAvailable; beside-prod boots die at plan time with `ExpertPlanError:
+host holds 5113 of 12096 swappable experts`). Save `podman logs glm-probe`
+to a file BEFORE teardown — a `podman logs | grep | head` pipeline died once
+under pipefail via SIGPIPE before log-save, losing the capture. The abort
+path must restore prod (unmask + daemon-reload + reset-failed + start),
+never bare-exit after stop+mask. Launcher knobs: `DFLASH_PATCH=1` (mounts
+the patched model.py), `DRAFT_MOUNT=1`, `SPECJSON_FILE=…` ({"method":
+"dflash","model":"/draft","num_speculative_tokens":8}) — note the launcher's
+echo prints `patch=$PATCH_MOUNT` (the COMPILE knob), not DFLASH_PATCH.
+Boot timeline: EAGLE3 gate fires ~4 min in at worker init; drafter
+torch.compile is the frontier (~380 s).
+
+**Open telemetry discrepancy:** the contaminated window read 1.5–2.1 tok/s
+with acceptance 1.00–1.25; Frank live-measured ~18 tok/s decode / ~370 tok/s
+prefill. Reconcile with a clean single-request measurement before trusting
+either.
+
+Related: [[dankserver-qwen38-fp8-expert-offload-tuning]] (the same fork, the same
+hardware — its conclusion was drawn from throughput arithmetic and is superseded
+by this page's kernel-time measurement),
+[[dankserver-frontier-moe-feasibility]] (the earlier fit analysis),
+[[VLLM Index]], [[LLM/index|LLM fleet]].
+
+## The deployed shape
+
+| term | value |
+| --- | --- |
+| payload | `Intel/GLM-5.3-Flash-W4A16-AutoRound` — AutoRound **symmetric int4, group 128**, packing `auto_round:auto_gptq`, 34 shards + 2 extra safetensors, 113,074 tensors, ~170 GiB |
+| runtime | `docker.io/tcclaviger/vllm@sha256:da4029666…` (same digest-pinned image as the qwen cutover) |
+| engine | TP=2, `--enable-expert-offload --expert-precise-memory`, max-model-len 131,072 x 2 seqs, bf16 KV, util 0.98, NBATCH 2048, prefix caching |
+| MoE kernel | `R4dW4a16MoEMethod` → libr4d grouped W4A16 GEMM, 6.234 MiB per expert-rank slot |
+| expert cache | 14.51 GiB = **2,383 slots** per rank; 12,096 host-backed experts, **73.64 GiB pinned per rank** |
+| unit | `scripts/frank-primary.service` (root unit, `LimitMEMLOCK=infinity` for the pin) |
+| measured | **decode 9.9-10.0 tok/s**, **prefill 918 tok/s** (warm, 2.6k-token prompt), KV cache 138,561 tokens |
+
+```mermaid
+graph LR
+    A[Intel AutoRound W4A16<br/>sym int4 g128, auto_gptq] -->|quant_method auto-round| B[inc.py override to inc]
+    B --> C[inc_wna16_scheme<br/>_resolve_gptq_moe]
+    C -->|sym int4 g128 + gfx1201 + libr4d| D[R4dW4a16MoEMethod<br/>grouped W4A16 GEMM]
+    D --> E[ExpertLRUCache<br/>2,383 VRAM slots / rank]
+    D --> F[73.64 GiB pinned host rows / rank]
+    E --> G[decode step]
+    F -->|host misses<br/>1.05% of device time| G
+```
+
+## Why this payload and not another
+
+The W4A16 fast path on this fork is reachable **only** through the INC
+(AutoRound) quant method. `compressed_tensors` checkpoints of the same model —
+for example `dealignai/GLM-5.3-Flash-CYBERSECURITY-W4A16`, 182 GiB — resolve
+their MoE through the WNA16 oracle to `TritonWNA16Experts`, which **has no
+offload recipe**, and the engine refuses to serve at all: 170 GiB of experts
+cannot fit 64 GiB of VRAM, so offload is mandatory, not optional.
+
+The chain that makes Intel's checkpoint work, with the exact evidence:
+
+```
+[expert_plan.py]  quantization=inc            ← inc.py:364 overrides "auto-round"
+[inc_wna16_scheme.py:249]  INC GPTQ MoE on ROCm: R4dW4a16MoEExperts (libr4d grouped W4A16 GEMM)
+[r4d_w4a16_moe.py:682]     Using MoEPrepareAndFinalizeNoDPEPModular / R4dW4a16MoEExperts
+[expert.py:450]            Expert offload: 42 MoE layers staged; 2,383 cache slots in 14.51 GiB planned per rank
+[expert.py:413]            42 layers split: 0 VRAM-only experts, 2,383 swap slots, arena 14.51 GiB,
+                           host 73.64 GiB pinned, host rows prepacked on the GPU
+```
+
+`R4dW4a16MoEExperts` is what the offload machinery wraps: the offload plan is
+format-agnostic but requires the experts class to expose the hot/cold split.
+
+## The numbers, and the wall
+
+Measured on the deployed engine, 2026-09-30 (`bench_glm.py`, real prose, greedy):
+
+| depth | prefill | decode |
+| --- | --- | --- |
+| 2,621 tok prompt, run 1 (cold cache) | 259 tok/s | 9.88 |
+| 2,621 tok prompt, runs 2-3 (warm) | 918 / 885 tok/s | 9.91 / 9.41 |
+| 493 tok prompt (telemetry window) | 124 tok/s | 8.90 |
+
+Decode is flat across context and across every arena change — the signature of a
+per-step cost that is not KV- or residency-driven at the margin.
+
+### The measured profile, and the real wall
+
+**This section replaces an earlier explanation, which the measurement falsified.**
+The first reading of this page bounded the step with throughput arithmetic: 42 MoE
+layers x 8 experts x 6.234 MiB = 2.045 GiB per step per rank, 0.82 GiB of it
+missed at a measured 0.53-0.66 hit rate, giving "0.82 GiB in ~100 ms = 8.2 GB/s =
+the gather is the wall". That arithmetic was circular — it *assumed* the gather
+consumed the whole step — and it is wrong. The engine, asked directly, disagrees:
+
+A torch-profiler window (`--profiler-config`, `POST /start_profile` /
+`stop_profile`, 496 decode tokens at the deployed shape, rank0 trace parsed for
+device-activity time) gives the true split. Total device kernel time **46.18 s /
+496 tokens = ~93 ms per token** against a measured ~104 ms step (9.6 tok/s with
+the profiler attached):
+
+| ms | % | launches | kernel |
+| --- | --- | --- | --- |
+| 16,074 | **34.8** | 708,500 | `vectorized_elementwise_kernel` (22.7 us each) |
+| 4,602 | 10.0 | 42,000 | `r4d_gemm_moe_w4a16_nt_b16_m64` — the MoE GEMM |
+| 3,108 | 6.7 | 174,708 | `elementwise_kernel_manual_unroll` |
+| 2,943 | 6.4 | 138,252 | `wvSplitK_hf_sml_` — split-K GEMM (projections / head) |
+| 2,829 | 6.1 | 83,160 | router: `sbtopk::gatherTopK` + `warpMergeSortTopK` + `bitonicSortKVInPlace` (1,208 + 1,051 + 559 ms; the 1,051 row is not in the list below) |
+| 3,015 | 6.5 | 133,920 | `r4d_mhc_pre_finish/pre_partial/post` — hyper-connections |
+| 2,782 | 6.0 | 127,916 | inductor `triton_poi_*` / `triton_per_*` (the fusions that DO happen) |
+| **487** | **1.05** | 20,664 | **`r4d_moe_lru_gather_kernel`** — the accused |
+| 250 | 0.54 | 20,832 | `_moe_lru_tele_kernel` — the telemetry that measured it (TELEM was on) |
+
+So the gather is **1.05 %** of device time, and `r4d_moe_lru_fused_kernel` (the
+LRU *manager*, not the copy) is a further 1,599 ms / 3.5 %. Residency work and the
+PCIe link are not the story: 44 % of the budget is **pointwise glue** — bf16<->fp32
+conversions, adds, `pow`/`rsqrt` from fp32 RMSNorm, `launch_clamp_scale` quant
+passes, `reduce_kernel` means — spread over **~1,428 elementwise launches per
+token** (3,734 kernel launches per token in total). The GPU is busy 86 % of the
+wall clock, so this is real device work, not launch-gap overhead.
+
+Two consequences that kill three whole families of planned work:
+
+- **Spec decode cannot help this shape, and the MTP measurement was not an
+  accident.** Pointwise cost scales with the token count, so a draft batch that
+  verifies 4 tokens pays ~4x the glue for the same per-token output — which is
+  exactly why MTP4 measured "acceptance 3.76-4.90, decode unchanged". Same reason
+  the gather was never the bound.
+- **Arena size and hit rate are not levers.** Confirmed twice: `--expert-precise-memory`
+  (arena 13.08 -> 14.51 GiB) moved decode by nothing, and the gather is 1 %.
+
+## Fork work: the compile changelist (WIP, does not boot yet)
+
+The fork compiles its own flagship trunk — `vllm/models/qwen4_exp/amd/model.py`
+carries `@support_torch_compile` — and **not** its GLM trunk. The engine says so on
+every boot:
+
+```
+Cudagraph mode FULL_AND_PIECEWISE requires piecewise capture, but the loaded model
+provides neither a compiled submodule nor breakable CUDA graphs.
+Overriding to FULL_DECODE_ONLY.
+```
+
+Inductor fusion is therefore the only lever that touches the 44 % pointwise mass.
+Three patches are staged in `scripts/patches/glm53-compile-0*.patch`, each carrying
+its own evidence header, base/patched sha256 and the next blocker:
+
+| patch | target | fixes |
+| --- | --- | --- |
+| `-01-model-decorator` | `models/glm5next/common/model.py` | decorates `Glm5NextModel` (env `R4D_GLM_COMPILE`, default on) |
+| `-02-trace-safe-probes` | `model_executor/kernels/r4d_lib.py` | freezes 10 capability probes that do `os.path.exists`/`.so` lookups **inside the traced forward** |
+| `-03-fullgraph-optout` | `compilation/wrapper.py` | makes `fullgraph=True` opt-out (env `VLLM_COMPILE_FULLGRAPH=0`) |
+
+What the boots established, in order (each one ~7 min):
+
+1. **01 alone dies** on the first cold probe call inside the traced region:
+   `Dynamo does not know how to trace the builtin 'posix.stat'` —
+   `r4d_mhc_supported -> r4d_lib.r4d_mhc_enabled -> available -> os.path.exists`.
+   **`functools.lru_cache` does not protect these probes**: dynamo re-traces the
+   wrapped body to build its guard even when the cache is warm (it warns about
+   exactly this), verified in-image — a *warm* lru_cache-calling function still
+   fails `fullgraph=True`, while a plain module-global read compiles.
+2. **01+02 get past it** (`libr4d capability probes frozen: 10 warm [...]` in all
+   four processes) and die on
+   `RuntimeError: Graph breaks are not supported with aot compile`, because this
+   build runs AOT compile by default and AOT refuses breaks.
+3. **01+02+03 with `VLLM_COMPILE_FULLGRAPH=0 VLLM_USE_AOT_COMPILE=0` traces the
+   trunk successfully** (`Dynamo bytecode transform time: 1.08 s`) and then dies on
+   the decode forward with `KeyError: 'in_proj_qkvbfg_a'` (`Glm5NextModel.forward`
+   line 748 calling the decoder layer at line 456). `in_proj_qkvbfg_a` is the KDA
+   layer's merged column-parallel projection (`kda.py:217`, called at :407) — a real
+   submodule attribute name, also present in the `load_weights` shard table
+   (`model.py:780-785`). Reading the full traceback (the reported line is inside
+   dynamo-rewritten code) traces the throw to dynamo's own name→module map:
+   `torch/_dynamo/output_graph.py` registers into `nn_modules` at :1259/:1384,
+   looks up with `self.nn_modules[node.target]` at :3000 (a bare-name `KeyError`
+   when absent), and documents stripping `_buffers[..]/_parameters[..]/_modules[..]`
+   from those names at :1225. So the compiler is failing to **resolve a submodule by
+   name** while tracing the GLM decoder layer; whether the root cause is the fork's
+   module tree or dynamo's registration is not yet settled. The next step is the one
+   torch itself suggests in that error — `TORCHDYNAMO_VERBOSE=1` for the internal
+   stack trace (and note the launcher resets `ENVARGS`, so it must be added inside
+   `serve_glm_autoround.sh`'s COMPILE_PATCH branch rather than in the outer
+   environment).
+
+4. **04 removes the fused-norm call from the traced region** (`R4D_GLM_FUSE_QKV_RMSNORM=0`:
+   the GLM MLA block stops calling the fork's fused-norm JIT kernel, whose kernel
+   object in `models/common/ops/fused_qk_rmsnorm.py` is bound *after* the function
+   that uses it — `_FUSED_Q_KV_RMSNORM_KERNEL` at line 223 vs its use at line 211).
+   **It removes a real break** (under `fullgraph=True` the fused call was the first
+   break dynamo reported; after this patch the report moves on to the workspace
+   acquisition) — but it is not by itself what blocks a boot, because with breaks
+   genuinely tolerated the blocker is the name-map `KeyError`. See the run history
+   below.
+5. **A launcher bug that made four boots misleading — worth knowing because the
+   failure kept moving.** The scratch launcher's compile branches were restructured
+   on 2026-09-30 and the `-e VLLM_COMPILE_FULLGRAPH=0` line was dropped, so four
+   consecutive boots mounted the wrapper patch while `_FULLGRAPH` still defaulted to
+   True. With `fullgraph=True` **every** break is an error, so those boots reported
+   whichever break came first:
+
+   | config | outcome |
+   | --- | --- |
+   | `fullgraph=False`, AOT off | breaks tolerated → dies later on the dynamo name-map `KeyError: 'in_proj_qkvbfg_a'` |
+   | `fullgraph=True`, AOT off | dies on the first break (cell → workspace → compiler-disable) |
+
+   So the reported breaks are real *as breaks*, but they are not what blocks a
+   break-tolerant compile — and an earlier version of this page blamed
+   `TORCH_LOGS=graph_breaks` for the difference. **That was wrong**; the flag only
+   logs. The launcher sets the env var again.
+
+**How to see the breaks (and its trap).** `TORCH_LOGS=graph_breaks` names each break
+with a user stack, and `VLLM_USE_AOT_COMPILE=0` is required because AOT compile
+hard-refuses breaks — but a fullgraph=True run reports the first break as fatal, so
+check which configuration you are actually in (`podman exec <ctr> printenv
+VLLM_COMPILE_FULLGRAPH`) before concluding a break blocks the boot. For the name-map
+KeyError there is a ready-made probe: put `/home/dank/dynamo-probe` on `PYTHONPATH`
+with `VLLM_DYNAMO_DUMP_ATTRS=1` (a `sitecustomize.py` — a `.pth` on `PYTHONPATH` is
+NOT executed, and the image's own `sitecustomize` is chained so its apport hook
+survives) and it prints, at the miss, the node target, the near-miss keys, and
+whether another live `OutputGraph` holds the key.
+
+**State of the changelist (2026-09-30, latest).** The trunk **traces**
+(`Dynamo bytecode transform time: 1.08 s`) and the boot dies on the dynamo name-map
+`KeyError: 'in_proj_qkvbfg_a'`. Established about it, in order:
+
+- It is raised **inside the traced `Glm5NextDecoderLayer.forward`** — the frame chain
+  is `profile_run → _dummy_run → execute_model → glm4_1v.py:2358 → the compile wrapper
+  → eval_frame.compile_wrapper → Glm5NextModel.forward:748 (layer(...)) → module
+  _call_impl → Glm5NextDecoderLayer.forward:456 → KeyError`. `TORCHDYNAMO_VERBOSE=1`
+  adds **no** internal stack trace, i.e. it is a plain Python KeyError in the traced
+  frame, raised in a callee dynamo **inlined** (the reported line is the `def`, which
+  is how inlined frames surface).
+- It is **not** raised from any of dynamo's four `nn_modules` lookup sites:
+  `/home/dank/dynamo-probe` (a `sitecustomize.py` on `PYTHONPATH`,
+  `VLLM_DYNAMO_DUMP_ATTRS=1`) wraps three of them and stayed silent across two runs
+  with the probe confirmed installed in both the API-server and worker processes.
+- It is **not** a string-keyed lookup of a literal: `in_proj_qkvbfg_a` appears in the
+  whole tree only as the KDA layer's attribute (`kda.py:217`, used at `:407`) and in
+  the `load_weights` shard table (`model.py:780-785`). The four
+  `proj_name = prefix.split(".")[-1]` sites in the quantization utils do compute
+  exactly this bare key, but every one of them is guarded (`if proj_name not in
+  fused_mapping: return`), so they cannot raise it. The pluggable-layer registry is
+  keyed by class name, not module path.
+- It reproduces on **every** run of this configuration (5 of 5), and it is
+  independent of patches 04/05: with breaks genuinely tolerated, those remove breaks
+  that were never fatal in the first place — the KeyError is the only thing standing
+  between this changelist and a boot.
+
+**Next instrument, precisely:** the inlined caller has to be named. Two cheap ways:
+`TORCH_LOGS=+dynamo` on a boot and grep the log for the key (dynamo prints inlined
+frames with their source), or wrap the inlining candidates in that forward — the mHC
+CustomOps (`hc_pre` / `hc_post` / `hc_fused_post_pre`, `layers/mhc.py`) and the
+indexer op — with a probe that reports on `KeyError`. Until that names the caller, this
+is a dynamo-side name-resolution failure and not a fixable fork bug.
+
+Two verified facts bound the rest of the shape: `suppress_errors` does **not** cover
+graph breaks (only `fullgraph=False` does), and the GLM trunk can never satisfy
+`fullgraph` because its indexer metadata syncs to host
+(`glm5next/amd/sparse_indexer.py:263,419`: `int(...max().item())`) — both are dead
+fallbacks on the decode path that the fork author already replaced, so they are not
+the blockers they appear to be at first glance.
+
+### Two structural constraints on any further fork work
+
+- **No new HIP kernels are possible.** The image ships `r4d.so` as a binary with no
+  sources. The obvious second lever — the 9 % router — is gated
+  `num_experts in {128, 256, 512}` in `router/r4d_route_router.py` and the binary
+  contains only `r4d_moe_route_kernel<128|256|512>` instantiations. GLM has **288**
+  experts, so it falls back to the generic sort-based top-k and **cannot be moved to
+  the fused router without rebuilding libr4d**. Only Python/config/inductor changes
+  (and runtime-JIT Triton) are available.
+- **`--kv-cache-dtype fp8` is a dead end for capacity too.** The plan's KV reserve
+  is dtype-blind (1.76 GiB at 131,072 either way), so fp8 buys no arena and no
+  window; it only removes the only MLA decode path gfx12 has (trap 1 above).
+
+### DFlash2 for GLM — what the drafter actually is
+
+`incoai/GLM-5.3-Flash-DFlash2` is **not** a GLM-shaped drafter and needs no MLA/DSA
+port: its `config.json` declares `architectures: ["DFlash2DraftModel"]` and
+`model_type: "qwen3"` — 5 dense layers, hidden 4096, 32 heads / 8 kv heads,
+head_dim 128, sliding attention (window 2048), `is_causal: false`,
+`use_cache: false`, `dflash_config` = block 8, conv group 16, selector rank 256 /
+top-k 16, and **`target_layer_ids: [5, 14, 24, 33, 42]` against
+`num_target_layers: 45`**.
+
+The fork already maps that architecture:
+`registry.py:636 "DFlash2DraftModel": ("qwen3_dflash2", "DFlash2Qwen3ForCausalLM")`,
+and the spec method is `"dflash"` (`config/speculative.py:67`). So the port is
+**target-side only**: the GLM trunk must expose auxiliary hidden states at those five
+layer indices, the same interface `deepseek_v2.py` (also MLA+MoE) implements —
+`set_aux_hidden_state_layers()` / `get_eagle3_default_aux_hidden_state_layers()` plus
+capture at `idx in self.aux_hidden_state_layers` in the layer loop. `glm5next`
+implements neither.
+
+**The load was tried, and the gap is exactly one interface.** Payload fetched to
+`/home/dank/models/GLM-5.3-Flash-DFlash2` and hash-verified (`model.safetensors`
+2,342,169,800 B, sha256 `b038e1d9…` = the Hub's LFS oid). Booting the engine with
+`--speculative-config '{"method":"dflash","model":"/draft","num_speculative_tokens":8}'`
+accepts the config and the drafter and dies at worker init with:
+
+```
+RuntimeError: Model does not support EAGLE3 interface
+```
+
+So nothing about the drafter, the method name, or the payload is wrong — the GLM
+trunk simply does not implement the aux-hidden-state contract the dflash runner
+requires. The reference is in this same image, twice: `deepseek_v2.py`
+(`set_aux_hidden_state_layers` / `get_eagle3_default_aux_hidden_state_layers` at :1948-1951,
+capture at :1554 and :1579) and the fork's own `qwen4_exp` trunk. The contract is
+small:
+
+- `self.aux_hidden_state_layers: tuple[int, ...]` on the trunk, default `()`, set
+  by the wrapper's `set_aux_hidden_state_layers()`;
+- in the layer loop, `if idx in self.aux_hidden_state_layers: aux_hidden_states.append(<single-stream state entering layer idx>)`
+  (DeepSeek appends `hidden_states + residual`, with a TP all-gather fixup when the
+  rank holds fewer tokens than `positions`);
+- `Glm5NextModel.forward` returns `(hidden_states, aux_hidden_states)` when the
+  list is non-empty;
+- `get_eagle3_default_aux_hidden_state_layers()` supplies a default triple when the drafter
+  config omits `target_layer_ids` (this drafter provides `[5, 14, 24, 33, 42]`, so
+  the runner derives trunk indices `[6, 15, 25, 34, 43]`).
+
+**The GLM-specific unknown** is what "the single-stream state entering layer k"
+means for a trunk whose layers carry mHC multi-stream state: DeepSeek's
+`(hidden_states, residual)` pair has no counterpart here — GLM's decoder layer
+returns `(hidden_states, residual, post, comb)` with `hc_expand`/`hc_pre` mixing
+`hc_mult` streams, so the plausible answer is the pre-mHC single-stream input `x`
+(4096-wide, matching the drafter's `hidden_size`), not a post-layer value. That is
+a semantic choice to validate by acceptance rate, not by inspection.
+
+**Budget no speedup.** Per the profile above, a verification pass pays pointwise
+cost proportional to its token count — the same reason MTP4 measured flat — and
+MTP4's acceptance was already excellent (3.76-4.90). The port is worth doing as
+*enabling work* (it is the prerequisite for any spec decode on this endpoint, and
+it is small), not as a performance fix, and the honest expectation on this shape is
+flat decode with a prefill regression like MTP's until the pointwise mass is fused.
+
+**The port exists now: `scripts/patches/glm53-dflash-01-aux-hidden-states.patch`** (WIP,
+independent of the compile changelist — it patches the same file as compile-01, so
+compile-01 goes first if both are applied; base `008eeaa2…` → patched `bc9913a1…`). It
+implements exactly the contract read out of this image: `EagleModelMixin` on
+`Glm5NextModel`, capture of the single-stream `hidden_states` ENTERING each requested
+layer in the loop, the conditional `(hidden_states, aux_hidden_states)` tuple return, and
+the three `SupportsEagle3` member attributes plus both interface methods on the
+wrappers (capture shape from `deepseek_v2.py:1551-1585`; tuple consumption at
+`v1/worker/gpu/cudagraph_utils.py:621`).
+
+**The patch went through five versions, and the decisive discovery was where the
+instance actually lives.** The traceback's `set_eagle3_aux_hidden_state_layers(self.model)`
+is called on the model the runner LOADED — and for GLM-5.3-Flash that is
+`Glm5NextForConditionalGeneration` (the multimodal wrapper; the APIServer log names it),
+which holds the text model as `self.language_model = Glm5NextForCausalLM`
+(`glm5next/common/model.py:1113`). Patching only `Glm5NextForCausalLM` (v4) still failed
+with the identical error because the `isinstance` ran against the wrapper. **v5 fixes
+both wrappers** and is the first version to clear the gate (both TP workers log
+`Using Eagle3 auxiliary layers from config: (6, 15, 25, 34, …)`).
+
+The full evolution, each an instructive trap:
+
+1. **v1** hand-wrote the two methods the traceback names → identical error. The methods
+   are not the whole contract.
+2. **v2** inherited `SupportsEagle3` as a base (the image's own idiom — `laguna.py:717`,
+   `mimo_v2.py:834`, …) → dies at class creation with `Cannot create a consistent method
+   resolution order (MRO) for bases …`. GLM's wrappers already carry four/five
+   Protocol-derived bases, and simulation confirmed the ForCausalLM's five-base ordering
+   cannot take `SupportsEagle3` in *any* position.
+3. **v3** declared the members as class attributes (MRO-safe) — but only on
+   `Glm5NextForCausalLM` → identical error (the wrong wrapper, see above).
+4. **v4** added all three members (`supports_eagle3`, `has_own_lm_head`,
+   `has_own_embed_tokens` — a runtime-checkable Protocol's `isinstance` walks ALL
+   `__protocol_attrs__` *including base-Protocol members* via `getattr_static`, verified
+   experimentally in-image with stock typing) on `Glm5NextForCausalLM` → identical error.
+5. **v5** adds the three attributes to BOTH wrappers + the two methods on both
+   (`Glm5NextForCausalLM` unwraps to `self.model` — the `EagleModelMixin` trunk;
+   `Glm5NextForConditionalGeneration` delegates to `self.language_model`, the
+   `pixtral.py:498` pattern for exactly this wrapper shape) → **gate passes**.
+
+**The lesson generalizes twice over**: when a runtime error names an interface, read the
+interface's declaration first — and establish WHICH INSTANCE the check runs against
+before patching anything: multimodal wrappers (`…ForConditionalGeneration`) are loaded
+models, not pass-throughs, and the text model behind `self.language_model` is invisible
+to a check run on the wrapper.
+
+**Retired diagnostic (do not resurrect as-is): the `/probe` startup hook.** An
+`op_probe.py` + `sitecustomize.py` pair was mounted to wrap vLLM module forwards and
+torch's name-map lookups. It produced no evidence and caused a **fork bomb** (see the
+hazard note under experiment-window discipline). Use `TORCH_LOGS` for dynamo questions
+instead — it needs no startup hook.
+
+Note also that the KV-cap patch's predecessor arithmetic in
+[[dankserver-qwen38-fp8-expert-offload-tuning]] should be read the same way: its
+"gather-bound" conclusion came from the same throughput arithmetic that this page's
+profiler window has now falsified.
+
+### The KV-cache grouping journey: clearing the unifier, then the kpool tail
+
+With patch 01 v5 the gate passes and the boot moves to KV-cache layout
+construction — where the next two blockers live, both in
+`v1/core/kv_cache_utils.py`:
+
+1. **The page-size unifier refuses GLM's MLA family.** With a draft model the
+   spec dict gains non-MLA drafter attention specs and `HiddenStateCacheSpec`
+   entries, so `_get_kv_cache_groups_glm5_next` (which requires ALL attention
+   specs to be `MLAAttentionSpec`) bails to the generic grouping path — whose
+   `unify_kv_cache_spec_page_size` raises
+   `NotImplementedError: Layer language_model.model.layers.3.self_attn.indexer.k_cache: page size is not divisible… cannot be padded. Padding is only supported for non-MLA attention layers.`
+   MLA layers cannot be *padded* (sparse MLA indexes whole token rows) — but
+   they CAN grow `block_size` when the max page is an exact multiple. The
+   indexer's page is just `tokens_per_state x` the MLA page, so
+   **`glm53-dflash-02-kv-unify-mla-multiple.patch`** bumps the unified max page
+   to a multiple of every `MLAAttentionSpec` page before the loop: all MLA-family
+   layers then take the legal divisible block-growth branch, drafter + mamba
+   layers pad. The 2026-09-30 window proved it: unify passes, memory profiling
+   passes, torch.compile passes, and the boot reaches cudagraph capture — the
+   NotImplementedError is gone.
+2. **The kpool tail kernel pins its block.** Cudagraph capture then dies in
+   `kpool_ops.kpool_decode_update_and_maybe_write_cache_batched`:
+   `assert tail_kv_cache.shape[2] == pool_size`. The tail cache must view as
+   `[num_blocks, 2, index_kpool, head_dim]` (`KpoolTailSpec(block_size=index_kpool)`,
+   `Glm5NextTailCache` in `glm5next/common/attention.py:159`), but the tail's
+   page (8192 B) *divided* the raised max page, so the unifier's divisible branch
+   grew its `block_size` by the ratio and the view became
+   `[num_blocks, 2, index_kpool x ratio, head_dim]`. Verified not the hybrid
+   promotion: the log carries no "Hybrid KV cache manager is disabled" warning,
+   so `disable_hybrid_kv_cache_manager` is unset and
+   `unify_hybrid_kv_cache_specs` never ran. **`glm53-dflash-03-kv-unify-tailpin.patch`**
+   (= 02 + a `KpoolTailSpec` branch that pads `page_size_padded` instead of ever
+   growing its block) **verified 2026-09-30**: the assert is gone and the boot
+   captures the DFlash2 drafter's FULL cuda graphs on both TP workers
+   (`Capturing model for DFlash2 speculator...`). Supersedes 02 in the repo.
+
+3. **The capacity wall: the generic path cannot alias the mamba states.**
+   Past capture, the boot dies at `_check_enough_kv_cache_memory`:
+   `Available KV cache memory: 0.92 GiB` against `31.76 GiB needed at max len
+   65536` (32.91 GiB at 131,072 — nearly flat, a fixed intercept, not
+   per-token KV). Two coupled causes, read out of the image:
+   - **The glm5 tensor layout aliases the mamba states INTO the MLA pages**
+     (`get_kv_cache_config_from_groups` places every mamba layer at the MLA
+     offsets — that is why GLM-5.3 fits 32 GiB cards at all, and why the
+     expert plan's KV registry is calibrated at 13,600 B/token). On the
+     generic path the mamba groups claim their OWN blocks, so ~30 GiB of
+     unaliased state pages appears.
+   - **The plan's KV reserve (`expert_activation_registry.py:_glm53_kv_bytes_per_token`,
+     a flat 13,600 B/token) knows nothing about the drafter.** The engine caps
+     KV at the plan's reserve, so even a correct generic-path estimate would
+     be starved.
+
+   Learned en route: the engine AUTO-SETS the attention block size to
+   **2,432 tokens** (`interface.py:918: Setting attention block size to 2432
+   tokens to ensure that attention page size is >= mamba page size`), so every
+   attention group's block counts scale from that floor — a 16-token drafter
+   block would need 4x the shared `num_blocks`; with the global 2,432 all
+   groups align.
+
+   **`glm53-dflash-04-kv-groups-draft.patch`** (+ **`-04b-kv-registry-dflash.patch`**
+   for the registry) keeps the aliased glm5 pool and extends it: the grouping
+   partitions the hidden-state specs (`HiddenStateCacheSpec` is an
+   `MLAAttentionSpec` subclass — excluded before the exact-type
+   `type(spec) is MLAAttentionSpec` partition) and the drafter attention
+   specs; the drafter layers form their own uniform group (per-group pages,
+   NO unification — the MLA block size stays at the engine-set 2,432 and the
+   kpool guard `block_size % 512 == 0` is never stressed); the layout returns
+   a 10-tuple with `draft_group`/`hidden_groups`; the builder sizes ONE
+   shared backing allocation (`allocate_kv_cache` asserts all tensor sizes
+   equal) as aliased pool + appended draft/hidden tensors with their own
+   strides; the memory estimator adds the extras. Patch 04b extends the
+   registry: 13,600 + 4,096 (drafter 8 kv heads x 128 x K+V x bf16) +
+   5 x hidden_size x 2 B (the aux hidden-state cache) per token when
+   `speculative_config.use_aux_hidden_state_outputs` is set.
+
+   The pure logic is smoke-tested locally (CPU torch + the image's real spec
+   classes against synthetic GLM+draft+hidden spec sets): grouping shape,
+   layout recognition, builder block math and memory accounting all pass.
+   Real-fit arithmetic: per-token ≈ 58,656 B → reserve ≈ 4.15 GiB at max len
+   65,536 (arena shrinks ~13.5 → ~10.3 GiB, ~1,650 slots), num_blocks ≈ 29 x
+   2,432 tokens ≈ 70,500 ≥ 65,536.
+
+All of these mount over site-packages via the scratch launcher's
+`KVUNIFY_PATCH=1` knob (host `/home/dank/dflash-patch2/new/…`, patched shas
+`5521e50b…` for `kv_cache_utils.py` and `1a623d5c…` for
+`expert_activation_registry.py`; apply-verified against the image's pristine
+files). The window script's decision pipeline was also fixed for a SIGPIPE
+death (`head -80` on a `podman logs` grep pipeline killed one run before
+log-save and restore; the log is now captured to a file FIRST, then grepped).
+
+
+## Measured arms (all 2026-09-30, one boot each)
+
+| arm | decode | prefill | verdict |
+| --- | --- | --- | --- |
+| baseline int4, MAXLEN 131,072, util 0.92, NBATCH 2048 | 10.03 / 10.08 / 9.41 | 877 / 878 / 885 | qualified incumbent |
+| + `--expert-precise-memory` (arena 13.08 → 14.51 GiB) | 9.88 / 9.91 | 918 | **neutral** — proves residency is no longer the binding term |
+| + MTP4 spec decode (draft loads, acceptance excellent) | 9.64-11.16 | **255** | **rejected** — see below |
+| + MTP4 + explicit `--kv-cache-memory-bytes` (patched cap) | boots, 151,767-token KV | — | enabler for the MTP arm only |
+
+### Why MTP is rejected here (it is not that it fails)
+
+MTP **works** on GLM in this fork: the head loads, the architecture resolves to
+`Glm5NextMTPModel`, and acceptance is excellent —
+
+```
+SpecDecoding metrics: Mean acceptance length: 3.83-4.90,
+Per-position acceptance rate: 0.917/0.833/0.583/0.500 … 0.962/0.846/0.692/0.615,
+Avg Draft acceptance rate: 69-97%
+```
+
+— and it buys **nothing**: decode 9.6-11.2 tok/s against 10.0 without it, while
+prefill **collapses 918 → 255 tok/s**. The reason is structural to offload: a
+verification forward gathers the **union** of the batch's experts, so spec
+decode amortizes compute but not a residency-bound gather. It also costs 2.6 GiB
+of arena. On a compute-bound (fully-resident) shape this would invert; here it
+does not.
+
+## Served surface (verified 2026-09-30 against the live engine)
+
+| capability | state | evidence |
+| --- | --- | --- |
+| `/v1/models` | `glm53-flash`, max_model_len 131,072 | model list probe |
+| completion | yes | `bench_glm.py` runs above |
+| **tool calls** | **yes** — needs `--enable-auto-tool-choice --tool-call-parser glm47` | a `get_weather` prompt returns `tool_calls: [{"id": "chatcmpl-tool-…", "type": "function", "function": {"name": "get_weather", "arguments": "{\"city\": \"Seattle\"}"}}]`, `finish_reason: tool_calls` |
+| **vision** | **yes** | a 64x64 solid-red PNG is answered "The image appears to be a solid red color…" |
+| reasoning channel | **yes** — needs `--reasoning-parser glm47` | without it the thinking leaks into `content` (measured: content began with the answer, then "Let me verify: … = 391", then a bare `</think`) with an empty reasoning field; with it, a train-time prompt answers in `content` ("**Step 1: Add the hours.** 14:20 + 3 hours = 17:20 …") and the chain of thought lands in **`reasoning`** — the vLLM field name the fleet documents (llama.cpp endpoints use `reasoning_content`) |
+
+Without the tool flags the engine answers **HTTP 400** — `"auto" tool choice
+requires --enable-auto-tool-choice and --tool-call-parser to be set` — which
+breaks every tool-using agent on the endpoint. Both parsers are `glm47`, the
+newest GLM dialect the fork ships (`glm45`/`glm47` exist for both).
+
+## Fork traps found (for the engine authors)
+
+1. **`--kv-cache-dtype fp8` is illegal on gfx12 with this model — and it buys
+   nothing even if fixed.** The `ROCM_AITER_MLA_SPARSE` backend's
+   `_use_rocm_sparse_triton()` requires a non-fp8 KV cache, and that rope-free
+   Triton path is the **only** MLA decode available on gfx12: aiter's MLA decode
+   op is registered under `@if_aiter_supported`, which is gated on **CDNA > 2**
+   (`is_aiter_found_and_supported()` → `get_cdna_version() > 2`). With fp8 KV
+   the MTP draft step demands an op that cannot exist here:
+   `RuntimeError: Worker failed with error ''_OpNamespace' 'vllm' object has no
+   attribute 'rocm_aiter_mla_decode_fwd''`. KV dtype therefore stays auto/bf16.
+   (`VLLM_ROCM_USE_AITER_MLA=1` does **not** help — the umbrella gate is the
+   CDNA check, not the env var.)
+   - **The capacity payoff is measured at zero.** The point of fp8 KV is fitting
+     more tokens in the same reserve, and this fork's reserve model is
+     **dtype-blind**: the fp8 boot printed the *identical* plan line as bf16 —
+     `KV reserve 1.76 GiB (13600 B/token x 131072 + mamba pages)`. The reserve is
+     dominated by the per-sequence KDA state (34 linear-attention layers) and the
+     DSA indexer's pools, not by the MLA KV bytes those 13,600 B/token barely
+     cover — so halving the KV dtype cannot move it.
+   - **What a fix would take, if it were ever wanted**: a new fp8 path in the
+     rope-free Triton kernels (`rocm_sparse_attn_prefill` / `rocm_sparse_attn_decode`
+     in `v1/attention/ops/rocm_aiter_mla_sparse.py`) that dequantizes pages
+     in-kernel, plus loosening `_use_rocm_sparse_triton()`. The fork's
+     `_r4d_fp8_paged_mqa_logits` shows the author already has an fp8 MQA-logits
+     path, but the sparse-MLA backend is not wired to it. Verdict 2026-09-30:
+     **do not spend effort here** — the payoff is capacity, and the reserve is
+     dtype-blind.
+2. **The plan's KV reserve under-models this model, and the cap enforces the
+   mistake.** The plan charges 13,600 B/token + a small mamba term, but GLM
+   carries a per-sequence KDA state (34 linear-attention layers) and, under MTP,
+   the draft's KV. `gpu_worker` then *caps* KV at that reserve, so a legal
+   `--max-model-len` becomes unbootable:
+   `ValueError: To serve at least one request with the model's max seq len
+   (131072), (2.07 GiB KV cache is needed, which is larger than the available KV
+   cache memory (1.74 GiB))`. Raising `max_model_len` only helps past ~440k
+   tokens (the cap's slope is ~13,845 B/token vs the need's ~12,700 + 0.52 GiB
+   fixed) — where the reserve would eat a third of the arena.
+   `scripts/patches/gpu_worker-kv-cap.patch` lets an explicit operator KV
+   request raise the cap; **it is not part of the deployed shape**, which boots
+   on the plan's own reserve.
+3. **A window is forced, not chosen.** The production `max_model_len 131072` is
+   what the reserve model requires for this payload (below ~92k the boot
+   refuses); it is not a capacity decision.
+4. **Expert prefetch remains poison** — unchanged from the qwen campaign
+   (36% precision, +40-51% decode from deleting it). No GLM arm re-opened it.
+
+## Operations
+
+```bash
+# rollback to the pre-cutover qwen FP8 + MTP4 shape (payload and R9V state
+# dirs are untouched on disk):
+git show 2339fae -- scripts/ | git apply   # or: git checkout 2339fae -- scripts/
+./scripts/deploy.sh dankserver --service frank-primary
+
+# payload inventory (34 shards + 2 extras, weight-map index):
+bash scripts/download-frank-primary.sh --verify-only
+
+# live telemetry (needs the profiler window; with_stack=false is mandatory —
+# with stack the stop path hangs and the trace buffer grows to tens of GB):
+podman logs frank-primary-glm 2>&1 | grep "expert_lru.py"
+```
+
+**Experiment-window discipline (learned the hard way, 2026-09-30).** This engine
+**ignores SIGTERM**: after `systemctl stop` its container sits in `podman ps` state
+`Stopping` for 20+ minutes *while still answering `/health` with 200*. That makes a
+liveness probe a lying oracle for a restore — twice a restore script printed
+`PROD HEALTHY` while `systemctl is-active` said `failed` and the runtime mask was
+still in place. The restore oracle must be a **conjunction**: `is-active` ==
+`active` **and** `is-enabled` == `enabled` **and** no `/run/systemd/system/<unit>`
+mask symlink **and** the container is owned by the unit and not `Stopping`. The
+repair recipe, all four steps needed:
+
+```bash
+sudo mv /run/systemd/system/frank-primary.service /tmp/stale-mask-N.service  # runtime mask SURVIVES stop and blocks start
+sudo systemctl daemon-reload
+sudo systemctl reset-failed frank-primary.service                            # a failed unit refuses to start
+podman rm -f frank-primary-glm                                               # SIGKILL; `podman stop` hangs on the same SIGTERM
+sudo systemctl start frank-primary.service                                   # then ~7 min to health 200
+```
+
+Also: **never let two experiment jobs overlap.** A later job's `podman rm -f
+glm-probe` cleanup ran while an earlier job was still mid-flight and deleted its
+container, so one whole round produced no evidence at all (`no container with name
+glm-probe found`) — check `proc://` for running jobs before opening a window.
+
+**Never put a diagnostic hook on `PYTHONPATH` without a marker guard — it is a fork
+bomb.** On 2026-09-30 a `sitecustomize.py` mounted at `/probe` (to wrap dynamo's
+name-map lookups and vLLM module forwards) produced a self-replicating chain of
+`python3` processes 19+ levels deep inside the probe container: `sitecustomize` is
+imported by *every* interpreter, the hook imported torch/vLLM at startup, something in
+that import chain spawned a helper python, and the child inherited `PYTHONPATH` (hence
+the hook) and spawned again. It was still alive 18 minutes later with the container
+still showing `Up`. Symptoms: `ps -eo pid,ppid,comm --forest` shows a deep ladder of
+identical interpreters, each child's PPID being the previous one. Guard (now in the
+hook): arm once per environment via a marker in `os.environ`, which children inherit —
+
+```python
+if os.environ.get("VLLM_PROBE_ARMED") == "1":
+    pass                       # an ancestor already armed it; stay inert
+else:
+    os.environ["VLLM_PROBE_ARMED"] = "1"
+    <install probes>
+```
+
+Kill recipe: `podman rm -f <experiment-container>` (the container cgroup owns every
+generation), then verify with a precise count over `/proc/<pid>/environ` for the marker
+variable — **never** pattern-kill an interpreter name, because the media stack on this
+host (romm, jellyfin, plex, glances) runs python in production.
+
+```bash
+# boot: ~7 min warm page cache, longer cold; health returns 503 while loading
+curl -s -o /dev/null -w "%{http_code}\n" http://127.0.0.1:7770/health
+
+# experiment windows (production stopped + `systemctl mask --runtime`, which
+# self-clears at reboot) ran from the host-side scratch launcher
+# /home/dank/serve_glm_autoround.sh: env knobs MAXLEN SEQS NBATCH UTIL KVBYTES
+# KVDTYPE SPEC SPECJSON_FILE TOOLPARSER REASONINGPARSER TELEM GCHUNKS GLANES
+# PROFDIR GLM_PATCH_MOUNT COMPILE_PATCH GATHER_PATCH DRAFT_MOUNT ENVARGS.
+# DRAFT_MOUNT=1 mounts /home/dank/models/GLM-5.3-Flash-DFlash2 at /draft;
+# SPECJSON_FILE passes a whole --speculative-config JSON (JSON with no spaces,
+# word-split by the launcher's unquoted $EXTRA).
+bash /home/dank/serve_glm_autoround.sh
+```
+
+Boot gate: the engine refuses a legal-looking shape in two measured ways — the
+KV cap above, and the profile-run activation peak (`TritonMLA` indexer scores:
+64 heads x 2,048 topk x NBATCH tokens x 2 B = 2.06 GiB at NBATCH 8192), which is
+what killed the first two boots at util 0.95.
+
+`PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True` is part of the qualified
+shape, not a tuning: the plan fills VRAM to its modelled headroom by design, so
+the default allocator's fragmentation fails the profile/graph-capture phase
+(measured on the deployed unit: "Tried to allocate 662.00 MiB. GPU 0 … 240.00
+MiB is free").
+
+The `deploy.sh` e2e gate also drives the admin panel's open-links spec, which
+walks every guarded vhost — including `http://192.168.0.112:8989/` on
+FrankBook-Pro. When that Mac mini is asleep the spec fails on a `page.goto`
+timeout and the deploy reports `DEPLOY INCOMPLETE` even though the engine is
+healthy; wake the host (magic packet to `de:6b:6d:c1:52:23`) or re-run
+`tests/e2e/run.sh` once it is up. Seen 2026-09-30: 32 passed, this one failed,
+and the WoL packet did not wake it (macOS "Wake for network access" needs
+toggling on the mini itself).
+
+## Rollback targets
+
+| state | where |
+| --- | --- |
+| qwen FP8 + MTP4 (2026-09-29 → 09-30) | commit `2339fae`, payload `/home/dank/models/Qwen3.8-Flash-Next-Uncensored-FP8` |
+| R9V radiance Qwen3.8-Flash-Next Q8_0 | [[frank-primary-r9v-radiance-qwen38-q8]], state dirs `/home/dank/r9v-state/{q8xl,uncensored}` |

serverAppConfigs-docs/LLM/frank-primary-llamacpp-qwen38-flash-next-q8.md

@@ -110,6 +110,32 @@ retires the lucebox user quadlet to dated home trash
 truly free, relinks + enables + restarts the root unit, and waits up to
 1800 s for `/health` (cold start page-ins ~175 GiB and builds the graphs).
 
+### Preflight the GGUF without downloading it
+
+GGUF metadata lives in the header and HF honours ranged GETs, so a few MB
+answers every sizing question before a 13 GB transfer
+(`Range: bytes=0-8000000`). Parse: magic `GGUF`, u32 version, u64
+n_tensors, u64 n_kv, then key/value pairs (string = u64 len + bytes; ARRAY
+= u32 elem type + u64 count; BOOL is 1 byte — the type hand-written parsers
+forget). Walk arrays by byte size even for a summary: a naive parser
+desynchronises on `tokenizer.ggml.tokens` and lands gigabytes off. Read:
+`general.architecture`, `<arch>.block_count`, `<arch>.context_length`,
+`<arch>.nextn_predict_layers` (MTP is a property of the FILE — the `-mtp`
+build carries `blk.<n>.nextn.*`; `--spec-type draft-mtp` then loads the
+head against the MAIN model's weights, and `--model-draft` would load the
+file twice), `<arch>.head_count_kv`, `<arch>.key_length`/`value_length`,
+`<arch>.full_attention_interval`, `<arch>.ssm.state_size`/`group_count`/
+`time_step_rank`, and for the projector `clip.projector_type` +
+`clip.has_vision_encoder`.
+
+KV arithmetic for the hybrid (16 × (3 Gated DeltaNet + 1 Gated Attention)):
+only full-attention layers take KV
+(`attn_layers × n_head_kv × head_dim × 2 × bytes_per_elem`; f16 = 2 B,
+q8_0 = 1.0625 B, q4_0 = 0.5625 B); DeltaNet state is a fixed per-slot
+buffer (~0.16 GiB/sequence at 48 layers/128 d_state/48 v-heads f32); the
+MTP draft context costs ≈ 4 KiB/token. `--parallel N` DIVIDES `--ctx-size`
+between slots — it does not multiply.
+
 ## Verification (2026-09-26 deploy)
 
 - `/v1/models` reports `frank-primary`.

serverAppConfigs-docs/LLM/frank-primary-lucebox-ds4.md

@@ -146,3 +146,70 @@ systemctl --user start frank-medium`) and stop/mask frank-primary. The
 _Filled in after the cutover: health, real completion, tool call, Open WebUI
 completion, measured prefill/decode, `systemctl --user is-enabled/is-active`
 state._
+
+## Dual-R9700 follow-up (2026-09-28 probe era)
+
+After this single-card runbook was retired, the DS41 engine got a dual-R9700
+attempt on the `feat/ds41-engine` branch (`main` rejects the checkpoint with
+`unsupported model architecture 'deepseek41'`). Findings that outlive the
+probe:
+
+**Budget both cards explicitly, or you waste half the box.** The engine
+sizes the streamed-expert cache LAST, after the model, so uncapped it absorbs
+every spare byte on the stream device — symptom: card1 at 33.0/34.2 GB
+looking busy while card0 sits at 14.2/34.2 GB; the busy card is holding a
+cache of things still streaming from NVMe. Pin all four:
+
+| env | card | meaning |
+|---|---|---|
+| `LUCE_EXPERT_BUDGET_MB` | hip:0 | primary hot experts (~11.96 MiB each) |
+| `LUCE_EXPERT_SECONDARY_BUDGET_MB` | hip:1 | secondary experts (needs `LUCE_DS4_MOE_TP*`) |
+| `LUCE_EXPERT_STREAM_CACHE_MB` | hip:1 | cap the auto-sized stream cache |
+| `LUCE_EXPERT_STREAM_DEVICE=1` | hip:1 | which card holds the slots and computes them |
+
+Working full-card shape: 21000 + 20000 + 8000 → card0 ~28.6/31.86 GiB,
+card1 ~27.3/31.86 GiB, resident experts 3,360 of 15,360.
+
+**Expert TP DOES work on gfx1201+gfx1201.** A retired quadlet recorded
+`LUCE_DS4_MOE_TP_INPROC` as "qualified for gfx1151+gfx1201, not
+gfx1201+gfx1201" — that is WRONG; peer access negotiates and the secondary
+tier comes up clean: `[deepseek4-moe-tp] enabled mode=in-process local=hip:0
+secondary=hip:1`. It is the only lever that materially raises resident
+expert count, and resident count is the entire bottleneck. Confirm every run
+with `expert ownership:` and `hybrid experts ready:` lines.
+
+**The placement file must live under the mounted models dir.** JSON
+`{"owners": [...]}`, 15,360 entries indexed `layer*384+expert`, values
+`primary`/`secondary`/`stream`. It is a ranking, not a promise: each device
+keeps what fits its budget and demotes the rest. Anywhere outside
+`/models/...` gives `cannot open expert placement` and a SILENT single-tier
+fallback — a silent-wrongness failure, not a crash.
+
+**V4.1 refuses layer split and layer-major prefill** — only `--ds4-prefill
+exact` (token-by-token) is legal on a 32 GiB card and `--ds4-fused-decode`
+is refused there. With ~22% of experts resident each token pulls ~187 cold
+experts × ~12 MiB; if `/proc/<pid>/io` `read_bytes` climbs into terabytes
+the model is larger than the box's RAM and you are in page-cache thrash at
+the NVMe ceiling — a capacity verdict, not a tuning problem. Low GPU util
+with a pegged link is the EXPECTED signature of PCIe-expert-stream-bound
+decode, not a broken run.
+
+**Harness traps, each of which produced a wrong number:** never `env -i` the
+podman call (rootless podman runs as `dank`; `env -i HOME=/root` makes
+PODMAN try `/root/.config` and die `stat /root/.config: permission denied`
+before the container exists — pass container env with `-e`); never put a
+large prompt on curl's argv (`Argument list too long` at ~128 KB — use
+`--data-binary @file`); size prompts by growing the file until `wc -c`
+reaches the byte target AND assert depth against the engine's own
+`usage.prompt_tokens` (nominal 4 chars/token was really 5.73; a chunk loop
+that appends a fixed ~152-byte line per nominal 200-byte chunk undershot
+30%); the server runs as the container's PID 1, so tear down with
+`podman rm -f`, never host `pkill`; drop page cache between arms
+(`sync; echo 3 > /proc/sys/vm/drop_caches`) or the first arm reads warm and
+every later one is a lie.
+
+Read the placement log — it is the whole diagnosis
+(`arch=deepseek41 layers=40 embd=5120 experts=384 used=6 ff=2304
+hash_layers=0` — hash_layers=0 is correct for V4.1; `engram: layers=[1,14]
+… 2 table(s) embedded (not mapped)` — the 189.13 GiB Engram tables are never
+mapped, so they do not evict the expert page cache).

serverAppConfigs-docs/LLM/frank-primary-r9v-radiance-qwen38-q8.md

@@ -40,7 +40,7 @@ The repo carries a **thin** wiring; the engine shape lives on-host:
 | Piece | File | Law |
 |---|---|---|
 | Unit | `scripts/frank-primary.service` | Root unit, `User=dank`, `LimitMEMLOCK=infinity`, `ExecStop=podman stop -t 120 frank-primary-r9v`, `Restart=on-failure` |
-| Env | `scripts/frank-primary.env` | Production overrides ONLY: bind 0.0.0.0, :7770, alias `frank-primary`, container `frank-primary-r9v`, GPU BDFs, dev-overlay adapter, `R9V_FULL_MUTABLE_CACHE=0`, sharp chat template |
+| Env | `scripts/frank-primary.env` | Production overrides ONLY: bind 0.0.0.0, :7770, alias `frank-primary`, container `frank-primary-r9v`, GPU BDFs, dev-overlay adapter, `R9V_FULL_MUTABLE_CACHE=0`, `R9V_PLE_READY_TIMEOUT=1800`, prefix caching ON, sharp chat template |
 | Serve | `scripts/serve-frank-primary.sh` | Layers setup state under repo env → execs the fork's own `scripts/launch.sh` via the docker→podman shim → `docker wait` |
 | Deployer | `scripts/deploy-remote.sh` `deploy_llm_primary` | Gates on setup state ready, image digest, PLE bytes, all six shard files + mmproj; MainPID false-green check |
 
@@ -140,9 +140,11 @@ edited in place — mutating it drifts the qualification.
   ssd-mmap residency).
 - Context envelope: `R9V_MAX_MODEL_LEN=131072`; harness `models.yml`
   carries `contextWindow: 131072`, `maxTokens: 36864` (compaction-gap
-  math: 70% × 131,072 = 91,750 + 36,864 = 128,614 ≤ 131,072).
-- Prefix caching OFF (`R9V_ENABLE_PREFIX_CACHING=0`), CACHE192 off — the
-  KV envelope at 131,072 is smaller than the uncensored 196,608 shape.
+  math: 70% × 131,072 = 91,750 + 36,864 = 128,614 ≤ 131,072). The KV
+  envelope at 131,072 is smaller than the uncensored 196,608 shape.
+- Prefix caching OFF at cutover (`R9V_ENABLE_PREFIX_CACHING=0`), CACHE192
+  off — superseded 2026-09-28: prefix caching is ON (see bench matrix
+  below).
 
 ## Prefill / decode bench matrix (2026-09-28, scratch port 7773)
 
@@ -183,9 +185,74 @@ prompt TTFT 44.2 s → **3.6–4.3 s** (9x), 131k 439 s → **6.9 s** (63x).
 Hit pattern: rounds 1–2 miss, round 3+ hit (identical-prompt and
 conversation-growth probes both). For the harness's resend-every-turn
 traffic that is a ~40–440 s TTFT win per steady-state turn; the one-turn
-miss costs nothing extra. **RECOMMENDED for production — pending Frank's
-go-ahead** (one env line; the launcher default is already ON, the q8 state
-opted out).
+miss costs nothing extra. **DEPLOYED 2026-09-28** (Frank's go-ahead): the
+repo env sets `R9V_ENABLE_PREFIX_CACHING=1` (commit `41e9189`), verified
+on :7770 (`--enable-prefix-caching` in the container argv, health 200,
+live completion answering).
+
+## Authoring a new profile (the qualification pipeline)
+
+Order matters at every step:
+
+1. **Package manifest** `packages/models/<pkg>/package.json` — target shards
+   + aux artifacts with sha256. Verify every target's byte count AND sha256
+   after fetch (`curl -fL -C -` per shard; never trust a fetch without `-f`).
+2. **Placement bootstrap** from GGUF headers: parse all shards' tensor
+   tables (a KV-only shard1 is normal for split parts; each part carries its
+   own tensor table). Per-expert full per layer =
+   (`ffn_gate_exps` + `ffn_up_exps` + `ffn_down_exps`)/512; rank slices
+   rank0 = full×416//640, rank1 = full−rank0. Hot counts =
+   floor(measured_budget / per-expert total). **Anchor budgets to the
+   MEASURED hot_bytes of the qualified sibling profile, never invented.**
+3. **Trim + validate**: `tools/trim_experts.py --hot-counts N,M`, then
+   `expert_budget.validate_manifest(m)` +
+   `expert_budget.expert_memory(m, slots, {cache_rank}, runtime=<runtime.json>)`
+   (runtime/catalog are keyword-only). Declared hot_bytes must match
+   computed or it raises.
+4. **PLE derive** (containerized `tools/prepare_ple.py` with
+   `--expected-bytes/--expected-type/--expected-shape`); setup skips
+   re-derivation when the file exists with matching size.
+5. **profile.env**: set values BEFORE `source dual-r9700/profile.env`
+   (set-first-wins; an empty string does NOT survive `:=`).
+
+The canonical bootstrap.json wrapper schema is exactly
+`{schema: "r9v.placement.v1", id, model_package, hardware, artifact_path,
+artifact_sha256, scope}` with `hardware: "dual-r9700-32gb-asymmetric"`; the
+wrapper `id` MUST equal the id in the profile.json placement descriptor
+(setup rejects mismatches: "does not match profile placement"). Never
+fabricate qualification metadata — use
+`scope: "Unranked routing bootstrap only; does not certify fit or headroom"`.
+
+GGUF parsing laws for the bootstrap: expert tensor names are
+`blk.0.ffn_up_exps.weight` — split on "." gives FOUR parts, layer number is
+parts[1] (a 3-part assumption crashes `int("")`); outer-KV value-type sizes
+`{0:1,1:1,2:2,3:2,4:4,5:4,6:4,7:1,8:str,9:arr,10:8,11:8,12:8}`; tensor byte
+extents from consecutive OFFSETS, never a type-size table (mis-buckets tails
+silently); the PLE Q8_0 tensor on this arch is shape (160, 320001536),
+54,400,261,120 B, an entire shard.
+
+Setup/start: `./r9v setup <alias> --state-dir <dir> --model-dir <dir> --image
+sha256:<digest> --local-image --reuse-from <sibling-model-dir>
+--accept-model-license` (aux artifacts reuse byte-identical files; F16 vs
+Q8 mmproj will not reuse). Start requires the SAME `--state-dir`; fixed
+placement via `R9V_EXPERT_MANIFEST_PATH` skips headroom/catalog/seed raises.
+The host-only doctor FAILs when another instance holds the cards (it wants
+3 GiB free per rank; production at ~30 GiB/card is a hard wall) — a second
+instance CANNOT coexist, so qualification requires a stop+mask window for
+the incumbent, then unmask+restart.
+
+**Qualification gate (unqualified → qualified):** first start runs the
+workload battery (PONG / tool_calls / vision / long-ctx), the PLE pinning
+proof (`VmLck` ≈ PLE bytes in `/proc/<container-pid>/status` when
+RESIDENCY_MODE=pinned), and a VRAM headroom snapshot (floor 3.3 GiB). If
+headroom is violated, lower hot counts via re-trim and re-derive
+MAX_EFFECTIVE. Only then record state `ready: true` and author the runbook.
+
+Remote-execution gotchas from the same windows: `nohup X &` inside an ssh
+command races ssh teardown — the script can die before writing a byte; run
+long remote scripts synchronously, or setsid + disown, or via a system
+unit. zsh login shells abort on unquoted `===` separators; grep globs need
+quoting through ssh+zsh layers.
 
 ## Ops
 

serverAppConfigs-docs/LLM/frank-small-llama-mimo-q5km.md

@@ -0,0 +1,173 @@
+---
+type: reference
+title: "frank-small on Dankclaw — llama.cpp, MiMo-V2.6-Distill-Qwen-9B Q5_K_M GGUF (retired)"
+description: "The retired 2026-10-02/03 frank-small runbook — the Q5_K_M rung at two wide unquantized-KV slots on the RTX 3090, superseded 2026-10-03 by the Q6_K file. Kept as the rollback reference: it carries the measured boot-pair table that prices every pool shape on this card, the slot-window contract, and the misbehaves tree."
+tags:
+  - llm
+  - frank-small
+  - dankclaw
+  - llama-cpp
+  - rtx3090
+  - mimo
+  - gguf
+  - kv-cache
+timestamp: 2026-10-03
+---
+
+# frank-small — llama.cpp MiMo-V2.6 Q5_K_M (RTX 3090, retired)
+
+> **RETIRED 2026-10-03** — superseded by
+> [[frank-small-llama-mimo-q6k-2x200704]]: the ~3.5 GiB this shape left free
+> bought the **Q6_K quant rung** (same runtime, same 2 × 200,704 pool, same
+> f16 KV), which measures 2,849 MiB free and 94.44 tok/s — 10 % slower
+> decode is the measured price of the higher rung on this bandwidth-bound
+> card. This file remains the rollback reference and the owner of the
+> measured boot-pair table.
+
+> Its own history: one quant step down from [[frank-small-llama-mimo-q6k]]
+> (2026-10-01), resized twice on 2026-10-02 (three contexts with vision, then
+> the 2–3 GiB-band pool), and on 2026-10-03 morning reshaped to **two
+> conversations of 200,704 tokens with the KV VALUES unquantized** —
+> *"2 streams with 200700 tokens context or more… if we have extra space on
+> the 3090, unquantize the kv."* That shape measured 20,530 used / 4,046
+> apparent free (the true `nvidia-smi` free was 3,571 — see *Shape*).
+
+## Identity
+
+|Field|Value|
+|---|---|
+|Endpoint|`192.168.0.143:7771`, served alias `frank-small`|
+|Runtime|llama.cpp **master** (qwen35 model support), CUDA build `sm_86`, at `/home/dank/GitPackages/llama-master/build-cuda/bin/llama-server` — unchanged across the quant swap|
+|Payload|`bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF` **Q5_K_M**, 6,876,124,704 B (6.40 GiB), sha256 `04cf46f2b5584ef922697d6c063f94879434fb86c467f1a09e52daf0cddaf35f` + the repo's `mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf`, 918,166,048 B, sha256 `ff348f3180a63188aa7285db85f550fe38acb61dd013c599eb8bad08d2cc2576` — both at `/home/dank/models/MiMo-Q5KM/`, fetched and byte- and sha256-verified by `scripts/download-frank-small.sh`|
+|Unit|root `frank-small.service` (repo `scripts/frank-small.service`, `User=dank`, relinked), env `scripts/frank-small.env`, serve `scripts/serve-frank-small.sh`|
+|Repo pins|`tests/frank_small_unit.bats` (stub-argv seam + the VRAM arithmetic), `deploy_llm_small` gates in `scripts/deploy-remote.sh`|
+
+## Shape and the KV contract
+
+`--ctx-size 401408 --parallel 2` — llama.cpp divides the pool across slots,
+so each conversation gets `n_ctx_slot = 200704` (boot log 2026-10-03:
+`n_slots = 2, n_ctx_slot = 200704, kv_unified = 'false'`). `CTX_SIZE` MUST
+stay divisible by `PARALLEL`; the bats seam enforces it.
+
+The KV contract is Frank's, as of 2026-10-03: **both K and V unquantized
+(f16)** — `--cache-type-k f16 --cache-type-v f16`. The asymmetric contract it
+replaced (f16 K + `q8_0` V) had existed to make three slots fit; with two
+slots the exact-value budget opens, and 4 KiB/token × 8 layers × 2 of extra
+V traffic is the measured ~2 % decode tax. The serve script still passes both
+flags explicitly — the asymmetric-era law that a quantized V requires
+`--flash-attn on` stays pinned by the seam.
+
+**Pool sizes on this stack come from boots, not formulas.** The compute
+buffers, CUDA graphs and vision tower all bill separately, so every shape
+change is priced from measured rows:
+
+|Shape|KV dtype (K+V)|Projector|`nvidia-smi` used|Non-KV (weights + buffers + projector)|
+|---|---|---|---|---|
+|2 × 262,144 (pool 524,288)|f16 + q8_0|none|20,190 MiB|**7,390 MiB**|
+|3 × 196,608 (pool 589,824)|f16 + q8_0|f16|22,972 MiB|**8,859 MiB**|
+|3 × 180,224 (pool 540,672)|f16 + q8_0|f16|21,684 MiB|~8,419 MiB|
+|**2 × 200,704 (pool 401,408)**|**f16 + f16**|f16|**20,530 MiB**|**~7,986 MiB**|
+
+The 32 KiB/token full-attention KV price (8 layers × 4 KV heads × 256 dim ×
+2 B × 2 halves) prices the deployed pool at 401,408 × 32 KiB = **12,544 MiB
+of KV**; the card is 24,576 MiB, so the shape leaves **4,046 MiB free**
+(measured steady-state — the KV pool is allocated whole at boot, so this is
+the ceiling). The headroom above the 2–3 GiB floor is deliberate slack, not
+an invitation: another 32,768 tokens of pool (1 GiB) would trade the last
+safety margin for window the harness compacts before reaching anyway.
+
+**Why 2 × 262,144 with f16 V does not fit either.** The native window pair is
+524,288 tokens = 16,384 MiB of KV plus ~7,986 non-KV = 24.4 GiB: under the
+floor by a hair. 200,704 is the largest 4,096-aligned slice that keeps the
+measured band.
+
+## Measured on this shape (2026-10-03, 2 × 200,704 + vision, f16 KV)
+
+|Check|Result|
+|---|---|
+|Boot to healthy|~5 s, `n_slots = 2`, `n_ctx_slot = 200704`, `loaded multimodal model` logged|
+|VRAM|**20,530 MiB used / 4,046 MiB free** of 24,576 (`nvidia-smi`, steady state)|
+|Decode, temp 0, no thinking|**104.84 tok/s** (128-token completion at ~500-token depth; the 3 × 180,224 q8_0-V shape measured 106.8 — ~2 % is the honest price of exact V)|
+|Prefill|1,650 t/s on the same shallow 506-token probe (rates are depth- and prompt-dependent; not comparable across shapes)|
+|Batch geometry|`--batch-size 2048 --ubatch-size 1024` — ubatch 2048 failed the scratch boot; 1024 boots clean|
+|Vision|projector loads at boot (`loaded multimodal model`); red-PNG probe passed on 2026-10-02. Caveat discovered on 2026-10-03: probes must use a >= 256 px image — a 10x10 px PNG undersupplies image tokens and the model hallucinates ("it's a video")|
+|Tools / thinking|unchanged by the reslice: native `tool_calls`, thinking in `reasoning_content`|
+
+## What the slot geometry costs
+
+The per-slot window is the whole price, in both directions:
+
+- **A prompt over 200,704 tokens is a hard 400**, so `~/.omp/agent/models.yml`
+  sets `contextWindow: 200704` for this endpoint — the harness compacts before
+  it reaches the ceiling instead of discovering it.
+- **frank-large's tallest pre-compaction prompt now fits here.** Compaction
+  for frank-large sessions runs on this endpoint (`compactionModel` in
+  `models.yml`); frank-large compacts at 70 % of 262,144 = 183,501 tokens,
+  which did **not** fit the old 180,224 slot and fits 200,704 with 17,163
+  tokens to spare. The 2026-10-02 window trim had restored the fit for
+  frank-primary's smaller 140,492-token trigger; the 2026-10-03 reslice widens
+  the fit to frank-large itself. Nothing routes compaction automatically —
+  the point is the ceiling no longer forbids it.
+- Two simultaneous generations are the slot ceiling; a third request queues
+  behind a released slot (the mechanism measured on the 3-slot shape).
+
+## Client contract
+
+- Thinking arrives in **`reasoning_content`** (`--reasoning-format deepseek`),
+  the field `models.yml` reads for this endpoint.
+- **Thinking is off unless the request asks**: the MiMo GGUF template defaults
+  `enable_thinking` false, so `models.yml` ships
+  `chat_template_kwargs: {enable_thinking: true, preserve_thinking: true}` in
+  `extraBody`. A raw client that omits it gets a direct answer — measured, and
+  not a fault. Trivial questions produce no thinking even with the flag on.
+- Tools: `--jinja` + server-side parsing, verified 2026-10-02.
+- **Vision is on**, through the projector file (`--mmproj`). Use the
+  OpenAI-standard `image_url` content part — that path answers correctly
+  ("Red." on a solid-red PNG). The legacy top-level `images` field is **not**
+  read by `/v1/chat/completions`: the same request sent that way answered
+  "Brown.", i.e. it generated from text alone. `/completion` takes `images`
+  but refuses a request without `messages` in this build, so chat is the only
+  vision path.
+- Setting `MMPROJ_FILE=` (explicitly empty) in `scripts/frank-small.env` drops
+  `--mmproj` and returns ~1.4 GiB of the card to KV — the documented text-only
+  opt-out. `MMPROJ_FILE` pointing at a file that does not exist is a boot
+  refusal, never a silent fallback to blindness.
+
+## No spec decode
+
+The GGUF has no MTP tensors (the same fact that made SGLang's NEXTN a 0.00
+ghost), and the cross-model DFlash draft
+(`/home/dank/models/DFlash-Qwen35/qwen3.5-9b-dflash-Q8_0.gguf`) is an
+optimization candidate, not a default — acceptance has to be measured on this
+traffic first. The bats seam fails if a draft flag ships unannounced.
+
+## Rollback
+
+|Target|How|
+|---|---|
+|Previous shape: 3 × 180,224 + q8_0 V (2026-10-02)|`scripts/frank-small.env` `CTX_SIZE=540672 PARALLEL=3 CACHE_TYPE_V=q8_0` (measured 21,684 MiB used, 2,441 free) and `contextWindow: 180224` in `models.yml` — choose it when three simultaneous conversations matter more than per-slot window or exact V|
+|Q6_K, same runtime|`scripts/frank-small.env` `MODEL_FILE=/home/dank/models/MiMo-Q6K/MiMo-V2.6-Distill-Qwen-9B-Q6_K.gguf` + `MMPROJ_FILE` in that dir — **this is the deployed rung since 2026-10-03** ([[frank-small-llama-mimo-q6k-2x200704]]): at 2 × 200,704 it fits with 2,849 MiB free, no pool cut needed (the older "+1.4 GiB, drop 49,152 tokens" estimate held only for the 3-slot shapes)|
+|Two 262,144 windows|`CTX_SIZE=524288 PARALLEL=2 CACHE_TYPE_V=q8_0` (measured 20,190 MiB used, 4,386 free **with q8_0 V** — with f16 V the same shape is 16,384 MiB of KV + ~7,986 non-KV and does not fit) and `contextWindow: 262144` in `models.yml` — choose it when a single conversation needs the native window|
+|Vision off|`MMPROJ_FILE=` in the env file; the deployer gate then has to be satisfied too, so point `LLM_SMALL_MMPROJ` at any real projector file or drop the gate in the same commit|
+|SGLang + INT8|[[frank-small-sglang-mimo-v26]] — venv and old unit/env/serve trio still in place; the INT8 payload was staged to `/home/dank/.AGENTTRASH/20261002-retired-llm-payloads/` on 2026-10-02, so restore it there or re-fetch with `old/download-frank-small-sglang-int8.sh`|
+
+## When it misbehaves
+
+- Port 7771 dark: `journalctl -u frank-small -n 50`; the serve script exits 1
+  naming the missing file if the binary or GGUF moved.
+- `No CUDA GPUs are available`: the 3090 vanished from the bus or the UVM node
+  is missing — `nvidia-modprobe -u -c=0` (the serve script runs it).
+- Requests queue behind each other: two concurrent long generations are the
+  slot ceiling by contract (2026-10-03 shape); a third waits for a slot to
+  free (the mechanism measured on the 3-slot shape: 1.7 s for three, 3.0 s for
+  the fourth).
+- `400 … exceeds the available context size (200704 tokens)`: the conversation
+  outgrew ONE slot. This is not pool exhaustion — the pool is 401,408 tokens
+  and the per-slot window is the binding limit. Shorten the prompt, compact,
+  or roll back to the 262,144 shape above.
+- Vision answers a color it never saw, or claims the image "is a video":
+  either the client sent `images` instead of an `image_url` content part, or
+  the probe image is too small to supply real image tokens (use >= 256 px;
+  see *Client contract*).
+- Slow decode after a CTX or quant change: check `nvidia-smi` — pushing KV out
+  of VRAM into unified memory halves the rate silently.

serverAppConfigs-docs/LLM/frank-small-llama-mimo-q6k-2x200704.md

@@ -0,0 +1,158 @@
+---
+type: reference
+title: "frank-small on Dankclaw — llama.cpp, MiMo-V2.6 Q6_K, 2 x 200704, f16 KV"
+description: "The active frank-small runbook — llama.cpp master (CUDA sm_86) serving bartowski's MiMo-V2.6-Distill-Qwen-9B Q6_K GGUF on the RTX 3090: two conversations of 200,704 tokens with fully unquantized f16 KV, vision through the bundled projector, and the boot-pair arithmetic that prices every pool change."
+tags:
+  - llm
+  - frank-small
+  - dankclaw
+  - llama-cpp
+  - gguf
+  - rtx3090
+timestamp: 2026-10-03
+---
+
+# frank-small — llama.cpp MiMo Q6_K, 2 × 200,704, f16 KV
+
+> **ACTIVE since 2026-10-03.** The third shape Frank bought in two days on
+> this card: the freed third slot paid for **exact KV values**
+> ([[frank-small-llama-mimo-q5km]] documents that step and remains the
+> rollback), and the ~3.5 GiB still free then paid for the **Q6_K quant rung**
+> — the same file this endpoint served on 2026-10-01, come back as an
+> upgrade rather than a compromise. The pool held at 401,408 (2 × 200,704);
+> only the weights moved up.
+
+## Identity
+
+| | |
+|---|---|
+| Host / device | Dankclaw (192.168.0.143), RTX 3090, 24,576 MiB |
+| Endpoint | `http://192.168.0.143:7771` — alias `frank-small`, `/v1` OpenAI surface |
+| Runtime | llama.cpp master, CUDA sm_86 build `…/llama-master/build-cuda/bin/llama-server` (`GGML_CUDA_FA_ALL_QUANTS=ON`) |
+| Unit | root systemd `frank-small.service` ([`scripts/frank-small.service`](../../scripts/frank-small.service)), `User=dank`, `Restart=on-failure`, `LimitMEMLOCK=infinity` |
+| Serve script | [`scripts/serve-frank-small.sh`](../../scripts/serve-frank-small.sh), tunables [`scripts/frank-small.env`](../../scripts/frank-small.env) |
+| Weights | `/home/dank/models/MiMo-Q6K/MiMo-V2.6-Distill-Qwen-9B-Q6_K.gguf` — bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF **Q6_K**, 7,793,710,624 B, sha256 `ef96d05a…` (qwen3_5 hybrid: 32 layers, 8 full-attention + 24 Gated DeltaNet, native 262,144, no MTP tensors) |
+| Projector | `…/MiMo-Q6K/mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf` — 918,166,048 B, sha256 `ff348f…` (hardlink of the Q5KM dir's copy, link-count-2 verified; the dir is self-contained) |
+| Fetch | [`scripts/download-frank-small.sh`](../../scripts/download-frank-small.sh) — `-f`, byte count + LFS sha256 verified, `--verify-only`; both files verified against the pinned sha on the host before this deploy |
+| Seam | `tests/frank_small_unit.bats` (argv assertions against a stub engine: payload path, slot geometry, KV pair, batch pair, projector opt-out) |
+
+## Shape and KV contract
+
+`--ctx-size 401408 --parallel 2` — llama.cpp divides the pool across slots,
+so each conversation gets `n_ctx_slot = 200704`. `CTX_SIZE` MUST stay
+divisible by `PARALLEL`; the seam enforces it. Both K and V are unquantized
+(f16) — the flags stay explicit because `--flash-attn` must not be lost if V
+ever goes back to `q8_0` (the serve script pins it, the seam pins the pin).
+
+**The Q6_K rung is a weights-side spend only.** File delta +917,585,920 B
+(+875 MiB) against Q5_K_M, same KV, same slot geometry, same batch pair.
+Priced from the measured Q5 boot at this shape (non-KV 7,986 MiB): the boot
+landed **21,276 MiB used / 2,849 MiB true free** — non-KV measured 8,732 MiB,
+inside the 2–3 GiB floor, so the pool stayed at 401,408 and the
+12,288-token step-down ladder was never climbed. Quote `nvidia-smi`'s `free`
+field, never `total − used`: the driver holds ~450 MiB of reserve the
+subtraction hides (that error once overstated this card's slack by a
+half-GiB).
+
+**Quant ceiling, evaluated and closed 2026-10-03 (Frank: stay here).**
+`Q8_0` is the only rung above this one on the Hub (9,545,979,424 B = 9,104
+MiB, +1,671 over Q6_K): at the live pool it would leave 1,178 MiB free —
+under the floor — and the calibrated bandwidth law (it predicted this file's
+94.4 t/s from Q5_K_M's 104.84 within 1 %) puts decode at ~72 t/s, a further
+−24 % for the near-lossless rung. Fitting it was priced two ways (trim the
+pool to 2 × 182,272 — which also breaks frank-large's 183,501-token
+compaction-prompt fit by 1,236 tokens — or revert V to `q8_0`), and both
+were declined. Re-open only on evidence the Q6_K residual matters.
+
+**Pool prices on this card come from boots, never formulas** — the measured
+boot-pair table lives in [[frank-small-llama-mimo-q5km]] (*Shape and KV
+contract*) and still prices every rollback shape.
+
+## Compute-buffer geometry (2026-10-05)
+
+`BATCH_SIZE=4096 UBATCH_SIZE=2048` — both doubled from 2048/1024, and the trade
+measured rather than assumed, because the predecessor shape recorded **"ubatch
+2048 failed the scratch boot"** and this card has less free space than it did.
+
+**That failure was a different shape, and it did not reproduce here.** The
+service itself is the bench (one card holds one server, so no scratch port):
+the old rung was re-booted through a transient `FRANK_SMALL_ENV_FILE` drop-in
+and reproduced the 2026-10-03 boot **byte-identically** — 21,276 MiB used /
+2,849 MiB free — which makes the ubatch the only variable in the comparison.
+
+|Probe|ubatch 1024|ubatch 2048|Δ|
+|---|---|---|---|
+|Prefill, 8,928 tokens|3,972 / 4,077 t/s|**4,207 / 4,212 t/s**|**+4.6 %**|
+|Decode, 128 tokens at that depth|91.87 / 93.01 t/s|93.81 / 94.27 t/s|noise|
+|Prefill, 279 tokens|2,675.7 t/s|2,658–2,811 t/s|noise|
+|VRAM used / free|21,276 / 2,849 MiB|**21,846 / 2,279 MiB**|**−616 MiB**|
+
+Three readings:
+
+- **The gain only exists above the old ubatch.** A 279-token prompt is shorter
+  than either micro-batch, so it cannot see the knob — and it did not move.
+  That is the control proving the 8.9k delta is geometry, not drift.
+- **Decode is flat**, as every prior sweep on this card found. Ubatch is a
+  prefill knob.
+- **The price is 616 MiB out of a 2.8 GiB margin.** 2,279 MiB free is still
+  inside the 2–3 GiB band, at its bottom edge. `4096/1536` is the middle rung,
+  and the pool's 12,288-token (384 MiB) ladder in
+  [[frank-small-llama-mimo-q5km]] is the other way to buy room back.
+
+**Why 2048 is right HERE while 512 was right on the R9700 sibling: the curve is
+backend-specific.** The 2× R9700 layer split peaks at ubatch 512 and is 10 %
+off that peak by 2048 ([[frank-medium-batch-ubatch]]); on this CUDA card the
+curve climbs monotonically to 2048 (710 → 1,188 t/s on the 35B MoE sweep,
+1,944 → 2,423 t/s on the Ornith 8k sweep, both with decode flat). Never carry a
+ubatch verdict across backends.
+
+The logical `--batch-size` is VRAM- and throughput-neutral — two sweeps and this
+A/B agree — so 4096 costs nothing and buys nothing by itself; it is here because
+that is the pair the curve was measured at.
+
+## Measured on this shape (2026-10-03, Q6_K, 2 × 200,704, f16 KV)
+
+|Check|Result|
+|---|---|
+|Boot to healthy|model + projector resident in ~2.3 s (load log `0.02.27`), `n_slots = 2`, `n_ctx_slot = 200704`, `loaded multimodal model`, health 200; deploy e2e gate 33/33|
+|VRAM|**21,276 MiB used / 2,849 MiB free** of 24,576 (`nvidia-smi` `free`, steady state) — inside the 2–3 GiB floor with exact V AND the higher rung. Superseded 2026-10-05 by the batch doubling: **21,846 / 2,279** — see *Compute-buffer geometry*|
+|Decode, temp 0, no thinking|**94.44 tok/s** on the identical 506-token probe that measured 104.84 on Q5_K_M — **−10 %**. Decode on this card is weight-bandwidth-bound, and Q6_K streams ~10 % more bytes per token; the quant rung buys quality with decode rate, measurably|
+|Prefill|2,229.9 t/s (same shallow probe; rates are depth-dependent, not comparable across probes)|
+|Vision|512×512 solid-red PNG → **`Red`** (277 prompt tokens — the image really reached the tower). See the probe-size trap in *Client contract*|
+|Tools / thinking|native `tool_calls`, thinking in `reasoning_content`, thinking OFF unless `enable_thinking: true` (MiMo template default; `models.yml` ships it in `extraBody`)|
+
+## Client contract
+
+llama.cpp surface, live-verified on this model family: model id `frank-small`,
+native tools via `--jinja` (`tool_calls` in the response), thinking in
+**`reasoning_content`**, vision via `image_url` content parts. Two vision
+traps, both measured here:
+
+1. A top-level `images` field on `/v1/chat/completions` is silently ignored
+   and the model invents a color.
+2. **The probe image must be big enough to supply real image tokens.** A
+   10×10 px solid-red PNG came back "The image is a video, so I can't see its
+   content" — the engine itself warns at boot that Qwen-VL wants ≥ 1024 image
+   tokens, and a postage-stamp image lands one patch below usable signal.
+   512×512 answers `Red` in 19 tokens. Always probe vision with a ≥ 256 px
+   image; a wrong-color answer from a tiny image is the probe's fault, not
+   the endpoint's.
+
+`~/.omp/agent/models.yml` sets `contextWindow: 200704` — the PER-SLOT window,
+so the harness compacts before llama.cpp answers a longer prompt with a 400.
+
+## Rollback
+
+|Target|How|
+|---|---|
+|Q5_K_M at the same shape|`scripts/frank-small.env` `MODEL_FILE=/home/dank/models/MiMo-Q5KM/MiMo-V2.6-Distill-Qwen-9B-Q5_K_M.gguf` (+ `MMPROJ_FILE` in that dir) and the deployer default — buys back 875 MiB (3,571 free measured) **and the 10 % decode rate** (104.84 tok/s measured). Payload sha-verified on disk|
+|3 × 180,224 + q8_0 V|`CTX_SIZE=540672 PARALLEL=3 CACHE_TYPE_V=q8_0` (measured 21,684 used / 2,441 free **with Q5_K_M** — Q6_K does NOT fit that shape inside the floor; pair them explicitly)|
+|SGLang + INT8|[[frank-small-sglang-mimo-v26]] — payload staged to `/home/dank/.AGENTTRASH/20261002-retired-llm-payloads/`|
+
+## When it misbehaves
+
+Same decision tree as [[frank-small-llama-mimo-q5km]] (*When it misbehaves*):
+`No CUDA GPUs are available` ⇒ `nvidia-modprobe -u -c=0`; a 400 naming
+200,704 ⇒ the conversation outgrew one slot, not the pool; slow decode after
+a shape change ⇒ check `nvidia-smi` for KV pushed into unified memory; queue
+behind two long generations ⇒ the slot ceiling is two, by contract.

serverAppConfigs-docs/LLM/frank-small-llama-mimo-q6k.md

@@ -0,0 +1,110 @@
+---
+type: reference
+title: "frank-small on Dankclaw — llama.cpp, MiMo-V2.6-Distill-Qwen-9B Q6_K GGUF (retired)"
+description: "The retired 2026-10-01 frank-small runbook — the first llama.cpp + GGUF shape on the RTX 3090: two native-window sessions, the asymmetric KV contract (f16 K + q8_0 V) that motivated the move off SGLang, the VRAM fit at Q6_K, and what to check when it misbehaves. Superseded the same night by the Q5_K_M payload."
+tags:
+  - llm
+  - frank-small
+  - dankclaw
+  - llama-cpp
+  - rtx3090
+  - mimo
+  - gguf
+  - kv-cache
+timestamp: 2026-10-01
+---
+
+# frank-small — llama.cpp MiMo-V2.6 Q6_K (RTX 3090)
+
+> **RETIRED 2026-10-01 → 10-02** (one night): superseded by
+> [[frank-small-llama-mimo-q5km]], the same runtime and shape one quant rung
+> down. The Q6_K file stays at `/home/dank/models/MiMo-Q6K/` as the rollback
+> target.
+>
+> Original order (Frank, 2026-10-01 night): move this endpoint to the GGUF
+> path for **asymmetric KV** — unquantized K + q8_0 V — which only llama.cpp
+> exposes per-tensor. That contract is what the current deployment kept.
+
+## Identity
+
+|Field|Value|
+|---|---|
+|Endpoint|`192.168.0.143:7771`, served alias `frank-small`|
+|Runtime|llama.cpp **master** (qwen35 model support), CUDA build `sm_86`, at `/home/dank/GitPackages/llama-master/build-cuda/bin/llama-server`|
+|Payload|`bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF` **Q6_K**, 7,793,710,624 B, sha256 `ef96d05a2ddf2cbb450d1af1ac3860ec769d3575bafa70692ee5609bda3fad7d`, at `/home/dank/models/MiMo-Q6K/`|
+|Unit|root `frank-small.service` (repo `scripts/frank-small.service`, `User=dank`, relinked), env `scripts/frank-small.env`, serve `scripts/serve-frank-small.sh`|
+|Repo pins|`tests/frank_small_unit.bats` (stub-argv seam), `deploy_llm_small` gates in `scripts/deploy-remote.sh`|
+
+## Shape and the KV contract
+
+`--ctx-size 524288 --parallel 2` — llama.cpp divides the pool across slots,
+so each session gets `n_ctx_slot = 262144`, the model's full native window
+(boot log confirms `n_slots = 2, n_ctx_slot = 262144, kv_unified = 'false'`).
+
+Frank's KV contract: **K unquantized (f16), V `q8_0`** —
+`--cache-type-k f16 --cache-type-v q8_0`. Two hard dependencies ride on it:
+
+- **`--flash-attn on` is mandatory**: quantized V exists only on the FA
+  path. Without the flag the engine falls back silently and the contract
+  becomes a lie. The serve script pins the trio; the bats seam pins the
+  script.
+- The VRAM budget: attention KV lives in the 8 full-attention layers only
+  (4 KV heads x 256 dim). K f16 ≈ 16 KiB/token, V q8_0 ≈ 8.5 KiB/token
+  (34 B per 32-element block) ⇒ ~24.5 KiB/token over the 524,288-token pool
+  ≈ **12.6 GiB**. Weights 7.8 + KV 12.6 + compute buffers/graphs ≈
+  **20.9 GiB measured** against 24.5 GiB — the "room for checkpoints" the
+  order asked for. If graph capture OOMs, halve `CTX_SIZE`; never quantize
+  K to fix it.
+
+## Client contract
+
+- Thinking arrives in **`reasoning_content`** (`--reasoning-format
+  deepseek`) — the same field SGLang's qwen3 parser used, so `models.yml`
+  `reasoningContentField` did NOT move. Live-verified 2026-10-01 22:57.
+- **Thinking is off unless the request asks**: the MiMo GGUF template
+  defaults `enable_thinking` false; `models.yml` now ships
+  `chat_template_kwargs: {enable_thinking: true, preserve_thinking: true}`
+  in `extraBody`. A raw client that omits it gets a no-thinking answer
+  (measured — this is not a fault).
+- Tools: `--jinja` + server-side tool parsing, verified
+  `tool_calls` round-trip 2026-10-01 23:0x.
+- **Text-only now.** The GGUF carries no vision projector; the
+  in-checkpoint tower existed only in the SGLang safetensors. `models.yml`
+  `input: [text]`.
+
+## No spec decode (yet)
+
+The GGUF has no MTP tensors (same fact that killed SGLang NEXTN), and the
+cross-model DFlash draft (`/home/dank/models/DFlash-Qwen35/qwen3.5-9b-dflash-Q8_0.gguf`,
+Q8_0) is an optimization-loop candidate, not a default — `--spec-type
+draft-dflash` needs acceptance measured on this traffic first. The bats
+seam fails if a draft flag ships unannounced.
+
+## Build
+
+No ccache, CUDA 12.4 system nvcc with `--gcc-install-dir`-free gcc-13 host
+compiler (nvcc 12.4 rejects gcc ≥ 14):
+
+```
+cmake -B build-cuda -S . -DCMAKE_BUILD_TYPE=Release -DGGML_CUDA=ON \
+  -DCMAKE_CUDA_ARCHITECTURES=86 -DCMAKE_CUDA_HOST_COMPILER=/usr/bin/gcc-13 \
+  -DLLAMA_CURL=OFF -DGGML_CCACHE=OFF
+cmake --build build-cuda -j 16 --target llama-server
+```
+
+The SGLang venv (`/home/dank/GitPackages/sglang-cu128`) and the INT8
+AutoRound payload (`/home/dank/models/MiMo-V2.6-Distill-Qwen-9B-INT8-W8A16-AutoRound`)
+stay on disk as the rollback: repoint the unit's serve script and env from
+git history (`git show frank-large-gufo-q4kxl-20261001` era is BEFORE this;
+use the parent of the 2026-10-01 night swap commit).
+
+## When it misbehaves
+
+- Port 7771 dark: `journalctl -u frank-small -n 50`; the serve script
+  exits 1 naming the missing file if the binary or GGUF moved.
+- `No CUDA GPUs are available`: the 3090 vanished from the bus or the UVM
+  node is missing — `nvidia-modprobe -u -c=0` (the serve script runs it).
+- Requests hang at first token while `/health` is 200: both slots busy —
+  two concurrent long generations are the ceiling by contract.
+- Slow decode after a CTX change: check `nvidia-smi` — swapping KV out of
+  VRAM into unified memory halves the rate silently.