The template was staged in the working tree (then stashed by a peer for the
prometheus deploy) but never committed, so the live container kept serving
/patches/qwen-fixed-v22.3.jinja. Commit the repo-tracked template and flip
the quadlet: CHAT_TEMPLATE now points at it, mounted read-only at
/templates/qwen-sharp-v22.4.0.jinja. The entrypoint already passes
--chat-template on the shared exec path, so this applies to the FP8 shape.
Rollback: one env flip back to /patches/qwen-fixed-v22.3.jinja.
node-exporter is net=host, so its 192.168.0.143:9100 bind is a host bind,
not a published port — the bridge-networked Prometheus container cannot
reach its own host's LAN IP there (netavark DNATs published ports only).
The node-dankclaw target has been red by construction since the
observability cutover while the container reported healthy. Point it at
host.containers.internal:9100 (netavark delivers that to the host LAN IP,
where node-exporter listens) — the same rule frank-large and frank-small
already use. node-dankserver is a different machine and keeps its LAN
address. Update the cross-file contract test and the Targets table.
DFlash2 (100.0 t/s decode, 287.1 t/s burst-4) beats MTP self-speculation
(85.9 t/s, 179.7 t/s) despite a lower accept length (4.24 vs 5.66): the
small separate drafter steps faster than the target's own MTP layer applied
autoregressively. DFlash2 is the deployed shape; both swap-script paths now
default to dflash/7. Docs updated.
The entrypoint now reads the model's weight quantization from config.json
and serves the plain FP8 path (no MXFP4 patches, --quantization=fp8, MTP
self-speculation, ROCM_AITER unified attention with the R4D all-reduce)
or the MXFP4 patch-and-serve path. A model swap is now: change SNAP
(+ SPEC_METHOD/SPEC), restart.
- radiance-entrypoint.sh: quantization detection, gated MXFP4 env vars
and patch block, --quantization fp8 flag
- quadlet: SNAP -> Qwen3.8-27B-Uncensored-FP8, SPEC_METHOD=mtp, SPEC=8,
KV_MEM=0 (profile the shape, the documented plain-FP8 default)
- deploy-remote.sh: model-aware gates (read SNAP/SPEC_METHOD from the
quadlet; drafter gate only for dflash, r4d.so gate only for mxfp4)
- tests: quantization-aware drafter/spec and KV-pool tests
- hf-fetch.sh: generic, resumable, verified, idempotent HF fetch tool
- frank-medium-fp8.manifest: 17 files (10 config + 7 shards) with sizes
and LFS sha256
- shared/prometheus/rules/llm-fleet.yml: the llamacpp reuse recording rule
divided cached by prompt, but llama.cpp's prompt_tokens_total EXCLUDES
cached tokens (its own HELP line), so the ratio was unbounded — measured
77 on a fully-warm frank-large, meaning LlmPrefixCacheCollapse could never
fire on that side. Now cached / (cached + prompt), with the traffic floor
counting both terms too; the header documents the unit trap. Also fixed the
LlmEndpointDown description: frank-large's CIRU llama.cpp build DOES export
the llamacpp:* family (verified live), so the alert covers all three
endpoints.
- shared/grafana/provisioning/dashboards/llm-fleet.json: re-cut for two
namespaces. Every panel reads both (the endpoint picker selects); the
KV-pool-fill panel retired 2026-09-16 returned with
vllm:kv_cache_usage_perc; the latency p95 panel returned with the vLLM
histograms (TTFT/ITL/e2e); the alerts panel is scoped to the fleet's
Llm.* rules; the stale halogen-era description is gone. All 21 panel
expressions validated against the live endpoints.
- runbook + log + Observability: prefix caching measured, not inferred —
reuse happens at 1,648-token block granularity (attention block size padded
to the mamba page size); an identical 13,021-token prompt went 2.30 s cold
to 0.68 s warm with 9,888 tokens (76 %) reused, while a 2,521-token prompt
registered zero hits. Rule counts corrected to 6 alerts + 3 recording rules.
./tests/run.sh: 325 pass.
The radiance TP=2 cutover changed the metric namespace frank-medium exports
(llamacpp:* -> vllm:*) and left two scrape targets addressing host processes
by an address a container cannot reach.
- shared/prometheus/rules/llm-fleet.yml: two namespaces side by side. Added
LlmRequestsWaiting, LlmKvCacheSaturated (vllm-only gauges llama.cpp never
had), the job:llm_vllm_prefix_cache_reuse:rate5m and
job:llm_spec_accept_length:rate5m recording rules, and re-cut
LlmPrefixCacheCollapse / LlmSpecDecodeOff to OR both families with
per-family traffic floors so an idle endpoint cannot fire them. Every
expression validated against live Prometheus.
- shared/prometheus/prometheus.yml: frank-small -> host.containers.internal:7771
and frank-large scraped at host.containers.internal:7770. Both are HOST
processes on Dankclaw and netavark DNATs published container ports only, so
the LAN-IP targets were red by construction; frank-large's CIRU llama.cpp
build does export the llamacpp:* family, contra the old comment.
- tests/observability_quadlet.bats: the two seams now encode the new contract
(per-endpoint reachability by process type; every named series must be on the
measured list for its engine; engine-independent rules must name both).
- homepage docs.html/services.json: frank-medium sees pictures, frank-large is
CIRU Orca with a single seat (was halogen w4b, two seats, 200k).
- docs: radiance runbook is ACTIVE (deployed flags, three-arm battery, verified
capabilities, rollback, replica-era sections labelled); llama.cpp runbook
carries the SUPERSEDED banner; Dual-R9700-Cooperation states the measured
verdict; Observability records the namespace split and the DNAT lesson;
AGENTS.md topology/frank-medium/Model Config/Benchmarking Policy/runbook
index flipped; both indexes updated; work-log entry added.
Also repaired agent-pass damage: prose prepended above frontmatter openers,
duplicated runbook bodies, a literal tool-call token pasted into a runbook, and
an index destroyed by diff artifacts (restored from HEAD, flips re-applied).
Vault link check: 128 pages, 0 unresolved. ./tests/run.sh: 325 pass.
The conversation battery (scripts/bench-conversation.py, salted, same 10
prompts, 2026-09-20) measured three arms on the two R9700s:
arm I llama.cpp layer split 32:33 + DFlash2, Q8_0 geo decode 45.7 t/s, TTFT 0.72 s, burst-4 71.7 t/s
arm L llama.cpp tensor split 1:1 (RCCL), no drafter geo decode 30.5 t/s, TTFT 0.71 s, burst-4 54.3 t/s
arm R radiance vLLM TP=2 + DFlash2-FP8, MXFP4 geo decode 95.8 t/s, TTFT 0.21 s, burst-4 259.4 t/s
Radiance wins every axis, so it takes the alias. The board swap is what made
arm R possible: TP=2 was archived on 2026-09-16 because the second card sat
behind B650 chipset bridges with AtomicOpsCap: Routing-, and the ProArt
B850-Creator put both cards on CPU GPP root ports (Gen5 x8, Routing+, ReqEn+).
The R4D one-shot all-reduce needs no RCCL collective anyway.
- un-retire dankserver/frank-medium/: the TP=2 quadlet (banner rewritten —
it is the deployed shape now), radiance-entrypoint.sh, the healthcheck.
The replica pair and its affinity proxy stay archived; one process owns
both cards now.
- retire scripts/frank-medium.{service,env} + serve-frank-medium.sh and
bench-spec-ab.py to scripts/.AGENTTRASH/20260920-llama-layersplit/ as the
rollback path. build-llama-rdna4.sh and download-frank-medium-gguf.sh
stay: they are the tooling that rollback needs.
- deploy-remote.sh: deploy_llm_medium now gates on the image (pull once,
re-tag to the localhost name the quadlet resolves), the MXFP4 checkpoint,
the drafter, the pinned patch clone and the patched libr4d; retires the
llama root unit into the home trash instead of relinking it; then
install_quadlet + wait_container_healthy 1800s (first start compiles).
retire_vllm_medium_quadlets becomes retire_vllm_replica_quadlets — the
a/b/proxy trio only, since frank-medium is installed again.
- planner + test-changed map dankserver/frank-medium/** to frank-medium
and its seam; the scripts-trio mapping is gone with the files.
- tests/frank_medium_unit.bats rewritten for the container contract: boot
persistence of a GENERATED unit (no enable on this host's v1 generator),
TP=2 with the atomics-era NCCL workarounds kept out, the DFlash2 depth vs
the MXFP4 decode-kernel row band, the KV pin's coupling to the shape it
was calibrated at plus pool/VRAM arithmetic, the endpoint/tool/reasoning
contract, and the archive invariants.
- homepage card + docs updated, and they were stale twice over: they still
described the ajgazin Swift 35:30 / 393k payload, not the deployed shape.
Known regression, documented in the unit: vision. The MXFP4 checkpoint is
language_model_only and the launcher loads no projector, so images are a 400
here; the harness's vision role already routes to frank-small.
The warm-up request was a warm prefix-cache hit on the next run, so it
hid exactly what the arms were meant to differ on (prefill). Token
counts now come from the streamed usage of each job itself, and --salt
defeats cross-arm inheritance.
bench-conversation.py answers 'what does this endpoint feel like': ten real
jobs streamed once, TTFT split from decode tok/s, salted so no arm inherits a
warm cache, optional concurrent burst for capacity. Needed to compare the two
tensor-parallel runtimes the 2026-09-20 board swap unblocked.
dankserver-route-primary.sh pins eno1 (the fleet address, .235) as the
preferred default route after the board swap brought up eno2 with DHCP and a
lower metric; eno2 stays up as failover. nmcli device reapply, not connection
up, so the link this session rides is not torn down.
librccl DT_NEEDEDs librocm-core.so.1, so a GGML_HIP_RCCL=ON binary would not
exec from its bundled lib dir (measured: 'error while loading shared
libraries: librocm-core.so.1'). The layer-split build never pulled it in.
Layer split needs no collectives, so RCCL stayed OFF. The 2026-09-20 board
swap put both R9700s on CPU root ports with atomic routing, so a
tensor-parallel (row-split) runtime is now buildable: pass
GGML_HIP_RCCL=ON with a distinct BUILD_DIR beside the production one.
Reanalysis after the 2026-09-20 swap to an ASUS ProArt B850-Creator WiFi
NEO: both R9700s now sit on CPU GPP root ports (x8/x8 Gen5 bifurcation),
every bridge reads AtomicOpsCap Routing+, and both endpoints read
AtomicOpsCtl ReqEn+ (CAP_EXP+28.w = 0x1440) — the exact precondition
Dual-R9700-Cooperation's test plan named. TP=2 is therefore unblocked,
not proven: no collective has been dispatched since the swap,
GGML_HIP_RCCL is still OFF, and the 272 t/s "tensor" number on record
measured a butterfly fallback.
- Dual-R9700-Cooperation: short answer rewritten, re-measured 2026-09-20
topology table, rebuild-hypothesis door marked moot, probe plan is now
standing work.
- Dankserver System Specs: Motherboard section; GPU BDFs 0b:00.0 ->
07:00.0, iGPU 10:00.0 -> 11:00.0, kernel 7.0.0-31, SMT on, and the new
eno2 (192.168.0.247) second 5GbE port holding the low-metric default
route while eno1 keeps the fleet's static .235.
- frank-medium-llamacpp-dflash2: rationale and hardware row split past
from present; dated work-log sections stay as history.
- AGENTS.md + both indexes: no longer assert a wall that no longer exists.
Docs only; frank-medium stayed active and healthy throughout (verified
/health). ./tests/run.sh 331/331.
The frank-medium runbook's deployed-flags table still carried the
predecessor's values (35,30 / 393216 / 196608 / draft-mtp / BF16 mmproj);
updated to the deployed shape (32,33 / 327680 / 163840 / draft-dflash / f16
mmproj) from scripts/frank-medium.env. The halogen runbook still described
itself as 'the active frank-large runbook'; marked RETIRED 2026-09-19 with a
pointer to [[frank-large-ciru-orca]].
The CIRU launcher gates on hip/lib/libamdhip64.so + rocr/lib/libhsa-runtime64.so
under its runtime root and exits 2 without them; the build links the private
venv through its RUNPATH, so point those entries at the venv's unversioned
devel symlinks (idempotent ln -sfn). Without this the unit crash-looped for
the full 2400s health gate on first deploy.
Operator call 2026-09-19: switch the iGPU endpoint from
halogen-flash-server to jcbtc/Qwen3.8-Flash-CIRU-STRIX-Orca on the CIRU
v4.4.1 runtime with the QSA window hotfix (host build against the private
ROCm 10 SDK), one stream of the full 262144 trained window. The service
leaves the quadlet shape for a root unit (the same shape as frank-small and
frank-medium: a private-SDK host binary has no image to install and no
generated unit to dodge); the halogen quadlet is archived, the deploy
handler retires it on the host, gates on 7770, and its stray-run sweep
matches the model path so it cannot take frank-small down.
Operator call 2026-09-19: switch the target to
orcarouter/Qwen3.8-27B-Uncensored-GGUF Q8_0 (gated repo; the fetch script now
requires an HF token and verifies sha256, not just size), take the drafter to
the Q4_K_M rung of z-lab/Qwen3.8-27B-DFlash2-GGUF, cut the pool to two
163840-token conversations (327680, entirely inside the trained 262144
window), and balance the tensors across the cards (32:33) in place of the
35:30 x16-lean. The in-file MTP head stays one SPEC_TYPE change away.
The vLLM detour (a2b39b1, same day) is retired: the Q4_K_M llama.cpp setup
is the requested shape and the vLLM KV-quant options on sm_86 all measured
dead (fp8 hardware-impossible, int8 page-size unification, turboquant
mm-prefix). Root unit restored, vLLM quadlet/Dockerfile/fetch script moved
to .AGENTTRASH, runbook listed as same-day-reverted. Rollback target:
frank-small-vllm-gemma4.
int8_per_token_head: page-size unification fails (scale tensor breaks
the 256/512 head-dim ratio). turboquant_k8v4: backend validation fails
(no mm-prefix attention, which Gemma4's vision/audio path requires).
fp8: hardware-dead on sm_86. bf16 it is; history documented in
frank-small.env, seam pins auto + no backend pin.
The VLLM_ATTENTION_BACKEND env var does not exist in this vLLM dev
build (warns 'Unknown vLLM environment variable' and is ignored), so
the pin landed nowhere and Gemma4's hook force-selected TRITON_ATTN
again. Pass --attention-backend TURBOQUANT as a CLI flag instead.
Gemma4's config hook force-selects TRITON_ATTN for the mixed 256/512
head dims whenever no backend is selected, and TRITON_ATTN rejects the
turboquant KV dtype (measured: 'kv_cache_dtype not supported', crash
loop). The TURBOQUANT backend accepts any head size and the
turboquant_* dtypes, and the Gemma4 MTP speculator carries the
target's selected backend through to the draft layers.
int8_per_token_head keeps a separate fp32 scale tensor per (token,
head); Gemma4's hybrid head dims (SWA 256 / full 512) make the
per-token page bytes 4160 vs 8256, not an integer ratio, so
unify_kv_cache_spec_page_size raises NotImplementedError — nine
crash-loop restarts measured. TurboQuant k8v4 packs its block scales
inside the tensor, keeps the 2:1 page ratio, and has a dedicated
backend with an explicit Ampere path. 289,647 -> ~772k-token pool.
fp8 KV is hardware-dead on the 3090 (sm_86, no cvt.e4m3x2), so int8 is
the cheapest quant the TRITON_ATTN backend can run: pure ALU dequant,
same backend, no fp8 tensor cores needed. 289,647 -> ~579k-token pool,
two -> four full-context sessions. Quality verified against the bf16
baseline (needle-in-haystack at 213k + short battery) before the swap.
podman 5.7.0's v1 quadlet generator rejects both ContainerEntrypoint
and Entry; the key is Entrypoint= (pkg/systemd/quadlet/quadlet.go
KeyEntrypoint, mapped to --entrypoint).
The podman 5.x v1 quadlet generator on dankclaw rejects
ContainerEntrypoint with 'unsupported key in group Container', so the
unit was never generated and the deploy died at restart with 'Unit
frank-small.service not found'. Entry= is the v1 spelling.
Runtime swap on the RTX 3090 (port 7771):
- vLLM gemma-x86_64-cu129 dev build, digest-pinned, audio extra pip-installed
into localhost/vllm/vllm-openai:frank-small (Dockerfile, no COPY lines)
- google/gemma-4-12B-it-qat-w4a16-ct (10.26 GiB int4) + Google's bf16 MTP
assistant (845 MB) — the w4a16-ct assistant cannot draft in vLLM (drafter
path skips Marlin; acceptance 1.00), bf16 measures 2.80-3.56
- root systemd unit -> rootless Quadlet (dankclaw/frank-small/quadlet/);
manual CUDA device+lib mounts (no nvidia-container-toolkit on host)
- bf16 KV (fp8 hardware-dead on sm_86), 289,647-token pool, 2 concurrent
full-context sessions, CUDA graphs on
- native text+image+audio: the fleet's multimodal endpoint
- measured: pp@67k 1293 t/s, pp@213k 608 t/s, tg@213k ~13.6 t/s,
tg@85k ~17.6 t/s
- host driver 610.57.04 (Ubuntu 26.04); sm_86 is correct, no arch override
- planner/scoping: frank-small now selected by dankclaw/frank-small/ +
env/serve/download trio; old unit retired to scripts/.AGENTTRASH/
- docs: new runbook frank-small-vllm-gemma4.md; AGENTS.md, LLM index,
top docs index updated; old runbook moved to retired
--cache-ram caps the total bytes of saved prompt states in HOST RAM, and
alloc() refuses any single state larger than the whole cap. That refusal was
firing in production: 137 times in the two days from 2026-09-16, at 11.0-11.4
GiB per state, so every conversation past ~170k tokens was uncacheable and
re-prefilled from scratch on its next turn.
The state law at this shape is 68 KiB/token (16 full-attention layers x 4 kv
heads x 256 dim x 2 B x 2 for K/V, plus the MTP draft context's 4096 B), so
8192 MiB held ~120k tokens and 16384 holds ~240k: a FULL 196608-token slot
state (~12.8 GiB) now fits with ~3 GiB spare.
Host RAM only, allocated on demand and never pinned, so the VRAM fit is
untouched. New seam: the cap must clear one full slot's state and stay under
half the host's RAM (it fails on the old 8192 and passes on 16384). Also
corrects the adjacent claim that 0 is the build's default for
--ctx-checkpoints -- this binary's --help says 32.
Isolated-port (:7790) 2k-token decode with AVG rocm-smi util over the
request (peak util lies: a layer split touches both cards every token):
35:30 draft ROCm1 31.52 t/s util0 41% util1 98% <- deployed
36:29 draft ROCm1 31.49 t/s util0 43% util1 98%
35:30 draft ROCm0 31.52 t/s util0 42% util1 98%
36:29 draft ROCm0 31.48 t/s util0 43% util1 98%
The x1 card (ROCm1) stays the 98% bottleneck in every configuration.
Moving the MTP drafter to the x16 card does not flip the balance: the
per-step cost is the verify phase (both cards run the full target) plus
the logits, which are pinned to the card holding the last layers. A
target-layer move (36:29) buys 2 points on the x16 card and erodes its
VRAM margin (fit free 0.45 GiB) for no throughput change. The x1 card is
compute-bound, not PCIe-bound (~5.2 MB activation per ubatch on the link),
so saturating it costs no tokens. 35:30 is the balance; no split change.
A pending entry for frank-default-sightsound (FLM NPU endpoint retired
2026-09-11) survived in the dankclaw state store and made every --pending
run exit 2 at the dispatcher before the Playwright e2e gate could
execute. The pending pass now drops entries the target no longer allows
into ~/.AGENTTRASH and keeps going. Behavioral seam test added.
- services.json: frank-large/medium/small move from admin-only to the user
page with current descriptions (halogen w4b iGPU; 35:30 layer-split MTP
27B on the R9700s; Gemma4-12B QAT vision path on the 3090) and docs
links to /docs#models
- docs.html: static user documentation (access chain, model guide, media
how-tos, troubleshooting); served at /docs via the post-auth homepage
listener; shipped by deploy_homepage; footer + index link to it
- tests: pages seam asserts the docs contract (page, link, Caddy route,
deployer copy)
- vault: Homepage.md synced (docs landing shipped, cards reclassified)
The two R9700s sit on unequal PCIe paths: ROCm0 (03:00.0) is CPU-root-port
Gen4 x16 (~31.5 GB/s); ROCm1 (0b:00.0) rides the B650 chipset through
bridges negotiating Gen4 x1 (~1.97 GB/s). With the MTP drafter, output
layers and heavier compute buffers pinned to the last-layers card, live
rocm-smi read 98% busy on the x1 card vs 40% on the x16 card. Moving one
DeltaNet layer (34) to ROCm0 shifts ~0.43 GiB off the bottlenecked card;
the seam fit lands 30.59/30.01 with gap 0.58 GiB, inside floor and slack.
36:29 fails both gates, so this lever is spent.
signal-cli 0.14.7 parses an all-digit -a value as a phone number (the
numeric account name never resolves), and the message is a -m option,
not a positional (a positional is parsed as a second recipient).
Verified live: corrected invocation returns type SUCCESS.
Every webhook died on 'cannot access local variable _last_sent' because
the prune comprehension rebound the module global inside the function.
Prune in place instead.
The bridge-subnet-only pair dropped the host's own health probe
(source 192.168.0.143/127.0.0.1) and failed a healthy deploy. The
deployer now installs the full set: ACCEPT 127.0.0.1, ACCEPT the host
IP, ACCEPT 10.89.0.0/16, then DROP.
169.254.1.2 is not a local address on this host (netavark runs its
bridges in aardvark-dns' netns), so the relay cannot bind it — and a
loopback bind refuses container traffic. Probe-socket proof: a host
socket on 192.168.0.143:18094 answers the container's 169.254.1.2:18094
dial while the LAN IP refuses from the container. The relay binds
192.168.0.143:18093 and the deployer installs an iptables INPUT pair
(ACCEPT 10.89.0.0/16, DROP rest) so the unauthenticated endpoint stays
off the LAN.
A socket bound to 127.0.0.1 refuses container traffic that arrives via
the podman bridge: the first live webhook failed with 'dial tcp
169.254.1.2:18093: connection refused'. The relay now binds the
host-gateway endpoint — link-local to the podman bridge, off the LAN —
and the deployer's health probe follows it there.
- runbook: identity now 4x262144, audio verified, throughput table with
the OOM and negative-result rows, deployed flags updated to reality.
- AGENTS.md frank-small paragraph matches the deployed shape.
- log.md: the sweep entry.
- scripts/frank-small.env: my earlier line-range edit had clobbered
NGL=999 and FLASH_ATTN=on out of the file; the seams caught it. The
live service was unaffected because the script's defaults match the
declared values exactly, but the file contract is restored.
Suite: ALL TESTS PASSED (321), vault 125 pages 0 unresolved.
Swept on the 3090 at the deployed 4x262144 geometry, salted API probes:
- ubatch 512: pp@254k ~918 t/s, tg@250k 90-204 (noisy)
- ubatch 1024: pp@254k 962 t/s (+4.8%), tg@250k 257-302
- ubatch 2048: OOM — the MTP draft context needs ~3.3 GiB of pp compute
buffer against 2.2 GiB headroom; do not try.
- draft depth 4 + poll 100 decoded WORSE at 250k (230-257) than 3/50:
the longer verification round re-reads more long-ctx attention than the
extra accepted token earns. Kept 3/50, now with the evidence attached.
Idle VRAM 23,300 MiB; all ubatch-1024 variants idled identically
(compute buffers reserved at max graph size), so 8192 buys the deepest
prefill window for free over 4096.
The 393216/34:31/MTP deploy measured 30.78 GiB on ROCm1 against the
fit's 28.32 — the MTP draft buffers (one nextn layer's KV, 4096 B/pos,
plus draft buffers) came in 2.46 GiB above the estimate. Recalibrate
the seam's MTP constant to the measured 4.20 GiB total charge on card
1 and drop the floor 1.5 -> 1.0 GiB for this shape (ROCm1's 1.08 GiB
headroom is real; 35:30/36:29 fail card 0's floor at this pool, so
34:31 is the only split that fits 2x196k).
AGENTS.md and the runbook move to the MTP drafter: 34:31, 393216 pool
(196608 per conversation), the DFlash2 sidecar documented as a
one-line revert, and a dated part-2 section with the fit/measured
table and the recalibration.
Both were implicit build defaults (3 and 50); the throughput sweep needs
to move them without editing the exec line each time. Defaults preserve
current behavior exactly. Additive only.
DFlash2's sidecar (1.92 GiB + buffers on ROCm1) kept the 393216 pool off the table; the MTP head lives in the target file, so its cost is one nextn layer's KV (4096 B/pos) plus small buffers. Fit: ROCm0 29.77 / ROCm1 28.32 GiB, over the floor, 34:31 the balance point at this pool. n_max 3 is the measured MTP winner. The seam's fit is now drafter-aware, the slack is the deployed integer gap (1483 MiB), the ctx cap 393216 with the rope caveat (top 131k extrapolates on the base rope).
CTX_SIZE=1048576 / PARALLEL=4: every slot gets the model's full trained
window. Measured fit arithmetic on the 3090: 1 full-context slot idles
at 11,996 MiB, 4 at 22,346 -> 3,450 MiB per slot, ~8,546 MiB fixed.
A 5th slot projects 25.8 GiB past the 24,576 MiB card, and the KV pool
allocates upfront, so 4 is the ceiling. Explicit CTX_SIZE bypasses --fit
on purpose: fit maximises the TOTAL pool and would divide a smaller
window across slots at parallel>1. The slot-grid seam now pins both the
full-window division and the budget inequality so a future capacity
bump fails pre-deploy, not in the boot allocation.
Audio verified while measuring: the mmproj's audio encoder transcribes
input_audio parts verbatim (JFK WAV sample). Header comments updated.
AGENTS.md topology, Active Services, the CUDA-quant-KV lesson (f16 is no
longer why), and the runbook index. New frank-small-gemma4-12b runbook:
payload digests, the SWA/q8 arithmetic behind --fit's 262k window, the
MTP-sidecar + ngram-mod chain, the always-accepted reasoning_effort, and
the fit-probe error that looks like a fault at boot. The Qwen runbook and
both indexes carry the retirement; work log updated.
AGENTS.md topology row and section, and the runbook: the pool is 327680
positions (163840 per conversation over two slots), the split is 36:29 -- one
layer past the 34:31 balance at this context, on purpose, for headroom on the
drafter card (fit 29.19/26.04, measured 28.40/26.66 GiB at load, 3.46/5.20
headroom) -- and vision is on via the explicit --mmproj flag, because this
build has no autoload (the first deploy came up text-only and 500'd image
requests). The rope caveat is written down where the cap lives: the GGUF has
no scaling keys, so the top 65k of the pool extrapolates on the base rope.
The Qwen3.8-27B IQ3_S file carried its own MTP head; Gemma4's ships as a
separate assistant file (mtp-gemma-4-12B-it.gguf, arch gemma4-assistant,
4 nextn layers) that loads via --model-draft and shares the target's last
layers' KV. The draftless ngram-mod matcher chains behind it in the same
--spec-type value — two context-backed drafters cannot share one process,
so this chain is the only shape that carries both.
KV is q8_0 (16 KiB/token/layer vs 32 at f16) and CTX_SIZE is unset so
--fit sizes the pool against what the 7.4 GiB of payloads leave on the
24 GiB card. build-cuda was rebuilt with -DGGML_CUDA_FA_ALL_QUANTS=ON,
which quantized KV on CUDA requires. The three seams that pinned the Qwen
contract (MTP-in-file, the Qwen mmproj name, the f16 KV arithmetic) now
pin the Gemma4 contract instead, and the deploy handler's presence check
points at the new payload.
The 36:29/327680/vision deploy came up healthy and text-only: image requests
answered 500 'image input is not supported - hint: ... provide the mmproj'
with no mmproj line in the load log, so llama-server fb27a525 does not
autoload mmproj*.gguf from the model directory. Name the file, the way
frank-small's serve script already does, and pin the flag in the seam.
The 320k pool (163840 per conversation over two slots) costs ~4.3 GiB of KV,
and the BF16 projector (888 MiB) lands on ROCm0. Fit at the deployed point,
computed, not estimated: ROCm0 29.19 / ROCm1 26.04 GiB of 31.86 -- 2.67 / 5.82
GiB free, both above the 1.5 GiB floor.
36:29 is one layer over the 35:30 balance on purpose: at this context the
balance point is 34:31 (0.11 GiB gap), and the operator moved a layer to ROCm0
to buy headroom on the drafter card. The gap is 3.15 GiB by construction, so
the seam's slack is that gap rounded up (3225 MiB) with the trade written down,
and the ctx cap moves 262144 -> 327680 with the rope caveat (the GGUF carries
no scaling; positions past the trained 262144 extrapolate on the base rope).
Vision: NO_MMPROJ=0, the serve script now refuses to start without the
projector when vision is on (a missing file is a deploy error, not a silent
text-only server), the download script fetches and byte-verifies it, and the
deploy handler's preflight checks for it. The projector is already on the host,
byte- and sha256-verified against the Hub.
6.44 / 6.90 is the state straight after load; after the service has served
traffic it sits at 6.23 / 6.64, with and without requests in flight. Both are
~5 GiB above the floor, but the runbook should not imply a load-dependent
difference that the two readings do not actually show.
New page [[frank-medium-batch-ubatch]]: the sweep's method (one invocation per
cell, a VRAM sampler per cell, 64k prefill measured AT 64k depth), the full
batch x ubatch matrix, the depth-0 control, the deployed-shape VRAM table, and
the limitations — including the one that cost me two runs, that amdgpu does not
return ~26 GiB at once, so a reload after a kill fails with "unable to allocate
ROCm0 buffer" and reads exactly like a configuration that does not fit.
AGENTS.md: the batch lesson now carries the measured curve (the peak at 512, the
10 % loss at 2048, the irrelevant logical batch, the invariant decode) instead
of the "~3.3 vs ~1.7 GiB" figure, which was measured at the bench's shape; at
the deployed shape the asymmetry is 2.41 against 1.30. Fleet row and the
frank-medium section now say 35:30 and batch 6144x512.
Runbook: deployed flags, the split history with the ubatch-512 table, the fault
note, and the verified VRAM (6.44 / 6.90 idle, 6.23 / 6.64 under two concurrent
requests).
Dual-R9700-Cooperation: the live capability table (card 0 ReqEn+ against card 1
ReqEn-, x1 effective link on card 1) and an honest label on the M.2_2 riser
premise — it is ASUS's spec sheet, not a measurement, and CPU-attached does not
imply Routing+ on this board (00:08.3 is CPU-side and Routing-).
log.md: the round trip recorded, newest first.
The batch/ubatch sweep (scripts/bench-batch-ubatch.sh, 23 cells of a 28-cell
grid, 3 reps each) says the ubatch knob has an OPTIMUM, not a plateau:
ubatch 256 384 512 768 1024 1280 1536 1792 2048
tok/s 1046 1068 1193 1135 1152 1151 1165 1148 1078
The curve is remarkably reproducible -- 512 came out 1192.88 / 1193.23 /
1193.28 / 1193.88 across batch 2048, 4096, 6144 and 8192, and 2048 came out
1077.38 / 1077.54 / 1077.59 -- so this is a deterministic property of the
shape, not noise. The logical batch is throughput-neutral (2048..8192 all
within 2 % at a fixed ubatch) and VRAM-neutral (the compute buffers are sized
by ubatch alone). Decode is 15.88 tok/s at depth 65536 on every single cell.
So the deployed 6144x1536 was both 2.4 % slower than the optimum and 3.71 GiB
more expensive. Measured at the deployed shape (262144, 2 slots, DFlash2):
36,29 ubatch 1536 -> 4.71 / 4.91 GiB free (what was deployed)
36,29 ubatch 512 -> 6.01 / 7.32 GiB free (ubatch alone)
35,30 ubatch 512 -> 6.44 / 6.90 GiB free <- deployed
The ubatch move freed 1.30 GiB on ROCm0 but 2.41 GiB on ROCm1, because ROCm1
holds the output layers and the drafter's graphs -- so it shifted the balance
toward ROCm1 by 1.31 GiB and the split had to follow it one layer. 0.46 GiB is
one DeltaNet layer's worth of asymmetry, i.e. the granularity floor: no split
can be tuned finer than a layer, which is why 34,31 is not better.
Re-fitted the seam with both ubatch coefficients measured at this shape and
re-anchored it on the new deployed point (it reproduces 6.44 / 6.90 to
0.01 GiB). Every neighbouring split still fails it: 36,29 -> 1.85, 34,31 ->
0.94, 37,28 -> 3.24, 1,1 -> 3.03, 39,26 -> 6.03 GiB.
Also replaced the unit test's literal `TENSOR_SPLIT=36,29` assertion with the
shape it actually cares about -- two shares that add up to the model's 65
blocks, with ROCm1 taking the fewer -- so re-measuring the balance no longer
breaks a test about parallelism. The numeric contract lives in the seam test.
pcie-link-test.cu (CUDA, pinned buffers, cudaEvent timing, both services stopped):
pinned H2D 6.72 GB/s and D2H 6.60 GB/s, i.e. 85 % of the Gen4 x4 ceiling. Async
is no better than sync and pageable sits within 1 % of pinned, so these copies are
link-limited rather than staging-limited — the usual "pageable costs 2x" does not
apply on this host. A 3 MiB slab costs 474 us, essentially 3 MiB / 6.72 GB/s plus
~27 us of launch.
That settles the expert-streaming question quantitatively: 1.70 GiB of active
experts per token (A3B @ Q4_K_M, 75 GiB expert mass) is 272 ms of link time ⇒
3.7 tok/s, 4.6 with a fifth of the experts resident. The same bytes computed in
host RAM at 256 GB/s cost 7.1 ms — the link is 31x slower than the memory behind
it, which is why llama.cpp's -ot exps=CPU path measured 20.46 tok/s while an
engine that streams loses by 3-5x. The port under the card is x8-capable and the
card is x16-capable, so the x4 is the adapter; doubling it still leaves streaming
well behind.
Also records FreeToken's own constraint: it keeps experts in host RAM with an LRU
GPU cache (exactly this streaming shape) and loads only modelopt block-FP8
safetensors for qwen3_5_moe, so Q4_K_M GGUF is unloadable there regardless.
The harness is in the repo now, so the runbook says how to run it and what the
tool will not tell you:
- / splits devices within one configuration and , splits sweep values, so a
legitimate --tensor-split 36,29 would have benched one card at a time.
- there is no -c: n_ctx = n_prompt + n_gen + n_depth, so no bench can decide
whether a configuration's buffers fit beside the deployed 262144 KV.
- a test that cannot allocate exit(1)s the process, hence one invocation per
cell.
- a shell's memlock limit (8 MiB here) is not the unit's (infinity), which
fails the load for the wrong reason.
- no --spec-* and no --parallel: decode figures are drafterless and slots are
never exercised; -d is what puts KV in front of the measurement at all.
Operator instruction: purge the cache so no run can serve a stale payload. The two
live payloads were cleared and then re-fetched rather than left absent — deleting
them without a re-fetch means neither endpoint starts.
halogen's re-fetch proved the point: its sidecars had been stale *in place*
(overlay.hgn 2,477,677,120 -> 2,572,466,560 bytes, overlay-speed.hgn
2,383,306,048 -> 2,478,095,488) while the 115.55 GiB checkpoint returned
byte-identical. frank-small's 27B and mmproj re-fetched to Hub-matching hashes,
unchanged.
Trap recorded: podman refuses a bind mount whose source directory is missing
(statfs ... no such file or directory), so frank-large needs /models to exist
*empty* for its first-start fetch to run.
Q4_K_S against the recorded Q4_K_M numbers: +2.1/+1.6/+2.6 % (iGPU-only pp512 /
pp16384 / tg64), +2.7 % tg32@64k, +4.1 %/+2.0 % for the two layer-split configs,
~0 on both pp512@64k, host RAM bottoming at 10.0 GiB against ~5 GiB. A memory
lever with a small speed dividend. IQ4_XS (97,473,155,360 bytes, sha256
b4787265...d0574f9d) is pulled and verified for the next comparison.
Also repairs an AGENTS.md edit that landed in the wrong section: the peer
session's commits shifted that file between my read and my write, so the
frank-large text briefly split the Model Config section.
`local b=$1 ub=$2 dir="$OUT/b${b}_ub${ub}"` expands every argument of the builtin
before the builtin assigns any of them, so $b was unbound and set -u aborted the run
four seconds in - after the service was already stopped. Two statements instead.
Found by running it under systemd-run: it exited non-zero with no cells run, which is
exactly what the per-cell TSV is for.
Question it answers: within the headroom the deployment has, how large may the logical
batch and the micro-batch be before the activation buffers stop fitting — and what does
each size buy in prefill?
- One llama-bench invocation per cell. llama-bench exit(1)s the moment a test cannot run,
so a single infeasible ubatch would otherwise take every later cell with it, and
infeasibility is one of the things being looked for.
- A VRAM sampler per cell, covering the load. Batch/ubatch sizes the activation buffers,
so a cell that looks fast because its allocation is smaller than intended has to be
distinguishable from one that is genuinely fast.
- The service is stopped for the sweep and started again from a trap, because a bench
beside a live service starves the same GPU.
- Run it from a transient unit: the deploy's unit sets LimitMEMLOCK=infinity for the ROCm
staging buffers and a shell on these hosts only gets 8 MiB.
- llama-bench separates -ts/-dev with SLASHES and has no -c, --parallel or --spec-*: the
context is n_prompt + n_gen + n_depth, and the drafter cannot take part.
Checking whether a download was still running with pgrep -a curl printed the
whole argv, Authorization header included, into the session transcript. The
token is not in the repo and nothing was committed, but the practice is wrong:
secrets belong in a -K config file or the environment, not in argv, and no
pgrep -a should ever target a command carrying an auth header.
The note described DSpark as the deployed drafter and 40:25 as the split; both
were true for a few hours. It now leads with DFlash2 at 36:29 and keeps the
excursion as the record it is:
- the drafter comparison table with DSpark's REAL size (1,455,376,576, not the
1,359,284,737 that went into the fetch script), and why it lost: no sliding
window, so its draft KV spanned the whole context.
- the split's two moves as one story: the batch pushed it to 40:25, the drafter
swap pulled it back to 36:29, with both measurement tables.
- the layer-size law (0.43 GiB DeltaNet, 1.43 GiB attention) as the reason the
balance is measured and not interpolated.
- the verified-behaviour section now carries the LIVE numbers: 4.71 / 4.91 GiB
free (4.43 / 4.61 under load, 0.18 GiB gap), the DFlash2 init lines, and
draft acceptance from both slots (0.63 / 0.48 on slot 1, 0.39 on slot 0).
- the fault note scopes its own evidence: 39:26 was loaded, not exercised.
DSpark came off after a day. It was the better-looking file (1.27 GiB against
1.92) but it declares no sliding window, and the draft context is built with
n_ctx = llama_n_ctx(ctx_tgt) -- so its KV spanned the whole 262144-token window
(5.0 GiB) and cost ROCm1 four layers of split to carry. DFlash2's window bounds
its draft KV at 2048 positions (~40 MiB), which frees ~4.3 GiB on the
draft-carrying card.
The balance point moved with it, so it was re-measured at the deployed batch
(6144 x 1536, 2 slots, 262144) rather than interpolated:
39,26 2.42 / 7.20 37,28 4.28 / 5.34 35,30 5.14 / 4.48
38,27 3.85 / 5.77 36,29 4.71 / 4.91 34,31 6.57 / 3.05
36:29 is the minimum (0.20 GiB apart) and leaves 3.2 GiB above the operator's
1.5 GiB floor on each card, so the full window stays. One -ts step is one layer,
and qwen35's layers are not the same size -- 0.43 GiB on a DeltaNet boundary,
1.43 GiB on a full-attention one -- which is why the table exists.
Also in here:
- the fetch script pointed at the wrong file (DSpark) AND declared the wrong
size for it: 1359284737 against the Hub's 1455376576, 96 MB short, which made
--verify-only fail on a payload that was in fact correct. Repointed at z-lab
DFlash2 Q8_0, 2056414816, and both files' sha256 were checked against the
Hub's LFS oids (host == Hub for both).
- the seam's fit was recalibrated for a window-bounded draft KV and for the new
point; it now passes 36:29 and fails every other split measured, including
1:1. Its ubatch term is an extrapolation and it cannot model contiguity, which
the comment says out loud.
- the serve script's dead dspark-only tunables (n-min, p-min, p-split, poll)
are gone; the build's defaults apply.
The note still described DFlash2, one slot at 262144 and a 1:1 split -- all of
which the day's work replaced. It now carries:
- the DSpark swap with the table that matters (1.27 vs 1.92 GiB, block_size 7 vs
8 so n_max 6 vs 7, a Markov head instead of the candidate selector, and NO
sliding window) and the rule behind the sizing: the draft context is built
with n_ctx = llama_n_ctx(ctx_tgt), so a drafter without a window caches the
whole target context -- 5.0 GiB here, against DFlash2's ~40 MiB.
- the failure that rule produces when ignored: load succeeds, /health answers,
and the first draft decode dies with an OOM at 1:1.
- why the split moved a second time: tripling the batch costs ROCm1 about twice
what it costs ROCm0, with the 38,27 -> 41,24 -> 40,25 measurement table.
- the deployed flags, the 2-slot shape (n_slots = 2, n_ctx_slot = 131072), the
DSpark init lines, and balanced 29.98 / 29.98 GiB cards.
- honest scoping on the older measurements: the throughput, drafter-worth and
ngram tables were taken on DFlash2, and the agent-loop numbers on one slot.
- the 39:26 page-fault signature as a thing to watch, marked as NOT an OOM.
The operator's revised spec: batch x3 (6144/1536), parallel 2, a 1.5 GiB per-card
floor, and the context maximised inside it.
Measured at 262144 with the old 38:27 split, 3x batch was brutally asymmetric:
ROCm1 (output layers + the drafter's graphs) paid ~3.3 GiB where ROCm0 paid
~1.7, leaving the draft-carrying card at 0.08 GiB free against ROCm0's 3.78.
One -ts step is worth ~1.5 GiB at this batch, so the balance is a measurement:
41:24 overshot (1.47 / 2.38), 40:25 landed at 1.92 / 1.96 GiB free.
That balance is also what maximises the context -- an unbalanced split would have
to buy its headroom out of CTX_SIZE, and CTX_SIZE is already at the model's
trained window (262144), now split over two slots of 131072. Verified under two
concurrent 1832-token prompts: 200/200, 1.87 / 1.86 GiB free, n_slots = 2,
n_ctx_slot = 131072, tool calls intact, and DSpark drafting on BOTH slots
(acceptance 0.38/0.22 and 0.52/0.43).
8192/2048 does not fit at this context at all: its 2020 MiB pp buffer fails to
allocate on ROCm1, and 4096/1024 fits but leaves 1.11 GiB -- both recorded in
the env's sizing table.
The unit seam now models both cards (share + draft + ubatch-scaled buffers,
calibrated to the rocm-smi figures to within 0.002%), asserts the 1.5 GiB floor
on each, and asserts the two cards stay within 0.5 GiB of each other. Both
assertions were confirmed to FAIL on a deliberately bad config (ubatch 8192, and
a 1:1 split) rather than pass by construction.
Correcting a claim I made from a bad query: Q4_K_S does exist for this finetune.
`mradermacher/Qwen3.8-Flash-Next-Uncensored-i1-GGUF` ships the large quants as
`.partNofM` fragments — Q4_K_M 37.00+37.00+36.97 = 110.97 GiB (which is how the
deployed file was assembled) and Q4_K_S 35.00+35.00+34.09 = 104.09 GiB — while
the smaller rungs are single files. A tree listing filtered on `.gguf` makes
Q4_K_S invisible, which is exactly how I concluded it did not exist.
Also records the fetch trap that hid it for a moment: `curl -o FILE URL` exits 0
on 401/403, so a gated repo produced three 153-byte "shards" with identical
sha256 values and a cheerful DOWNLOAD_DONE in 20 s for 104 GiB. -f plus the
Hub's declared size plus the LFS sha256 is the standard the fleet's own
download-frank-medium-gguf.sh already meets.
The 38:27 skew folded into the DSpark swap already does what step 3 asks: the
draft-carrying card sits 2.08 GiB above the other (5.49 / 3.41 GiB free) instead
of the ~9 GiB a 1:1 split would leave.
Chasing an exact balance is a mistake worth recording: 39:26 measured tighter
(3.86 / 4.69 GiB free) but that server died on its eighth request with SDMA0
UTCL2 page faults in the amdgpu driver -- "ROCm error: an illegal memory access
was encountered" -> ggml_abort, in the cross-card boundary copy that is the
layer split's hot path. 38:27 has since served the whole e2e gate and 12 varied
requests with no fault, so the residual 2.08 GiB gap is the safe side of this
trade. Comment-only: no restart needed.
Step 2 of the operator's six. --ctx-checkpoints 8 held per-slot KV snapshots;
0 is what llama.cpp ships, and the fleet's clients open conversations fresh.
DSpark declares no sliding window, so its draft context -- built with
n_ctx = llama_n_ctx(ctx_tgt) -- caches the whole target window: 5 layers x 2 x 8
kv heads x 128 dim x 2 B = 20 KiB/position, i.e. 5.0 GiB at 262144, where
DFlash2's sliding_window 2048 cost ~40 MiB. At a 1:1 split that put ROCm1 at
33.9 GB of 34.2 GB and the first draft graph capture died inside
common_speculative_impl_draft_dflash::process with "ROCm error: out of memory"
-- after the model had loaded and /health had answered, which is why the deploy
looked like a crash-loop rather than a sizing failure.
Handing three target layers to ROCm0 restores headroom on both cards at the full
window. Measured at 262144: 28.52 GB / 30.71 GB, i.e. 5.30 GiB and 3.26 GiB free
(was 2.7 GB free on the draft-carrying card with 1:1 + DFlash2). The unit seam
now models the busier card as a skewed share plus the drafter's full-context KV,
calibrated to the rocm-smi figure.
AGENTS.md gains the durable lesson: read sliding_window out of the drafter's
GGUF header before sizing a spec-decode deployment.
The log mtimes plus each run's own duration pin it: the first 'invalid internal
status' in the user journal is 18:58:30, while the F bench (launched 18:54:31)
was still loading, and the preceding llama-cli placement probe had just spilled
3.2 GB of '> ' prompts. The note had said 'after the experiment'; that was wrong
and is now recorded with the timestamp it can be checked against.
I had written 'memory pressure from the 116 GiB iGPU runs' as the cause of the
pause-process death. journalctl -k records no OOM kill for that window and the
kernel says nothing else about it, so the claim was a guess wearing a finding's
clothes — in AGENTS.md, the x4 note, and the log. All three now state the
trigger is not established, and the note's 'co-tenants' patience' bullet no
longer asserts that causality either. AGENTS.md also gains the real lesson:
journalctl -k is the check, and an unknown cause does not need a story.
DSpark (Q8_0, 1.27 GiB) replaces z-lab DFlash2 (1.92 GiB): same 5-layer
block-diffusion shape, Markov head instead of the candidate selector, and
block_size 7, so the draft depth drops 7 -> 6 (llama.cpp asserts
n_max <= block_size - 1).
The serve script's dspark branch never passed --spec-draft-model (only the
dflash branch did), so dspark was not deployable as written; the model file
gate now covers both types. Download script, env and the unit seam track the
new file and its exact byte count.
The ledger number in the note was pre-reboot. Same model, same pool, no clamp —
but the host was measurably more fragmented when it came back, which is the sort
of thing that decides whether the next co-tenant fits.
Asked whether Vulkan on both devices unlocks the split. It does not: the
per-tensor design aborts byte-identically, because the gate is not the driver
mix but `--override-tensor` — `has_tensor_overrides()` is a required term of
the scheduler's `pipeline_parallel` flag (llama-context.cpp:433, the only place
in the tree that reads it), and that flag sets `n_copies` 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.
Chasing it exposed a harness artifact that invalidated the earlier conclusion:
llama-bench splits a device LIST with `/` and SWEEP values with `,`, so
`-dev CUDA0,Vulkan1` was two single-device runs all along (and `-ts 0.85,0.15`
two sweep values, i.e. device 0 takes everything — which is where the bogus OOM
came from). With the correct syntax two-device splits work in both flavours:
pure Vulkan -dev Vulkan1/Vulkan0 -ts 0.85/0.15 -> 27.55 tok/s @64k (12.8 GiB on the 3090)
cross-vendor -dev CUDA0/Vulkan1 -ts 0.15/0.85 -> 27.23 tok/s @64k, pp512 268.5 (15.0 GiB)
iGPU alone 25.77 tok/s @64k, pp512 231.8
+7 % decode, +16 % prefill — real but not a regime change, and it costs the
3090 ~15 GiB, which this fleet spends on its only vision endpoint. The design's
premise still fails: whole layers carry their experts (1.62 GiB each), the 12
attention layers are spread every 4th, and the 32.8 GiB PLE must stay on the
main device, so KV cannot be consolidated on the 3090.
Also records the evening's incident (a dead rootless podman pause process took
all container management down; `podman system migrate` panics in 5.7.0; reboot
repaired it) and the netavark host-address lesson from the hermes 502.
hermes-webui is Network=host, so 8787 is a bare host-netns bind, and netavark
DNATs only *published* ports. From Caddy's bridge netns 192.168.0.143:8787 and
10.89.1.1:8787 are both refused, while host.containers.internal:8787
(169.254.1.2) answers 200 — measured live 2026-09-16 from inside the Caddy
container. The dashboard's published 9119 answers on that same LAN IP, which is
what made the old upstream look sound.
The vhost comment asserted "192.168.0.143:8787 is open" from that bridge; it is
not, so the claim is replaced with the measurement, and the hermes-webui quadlet
comment carries the same correction. The seam guarding this invariant accepted
any numeric upstream, so the LAN IP passed it; it now pins the measured-working
form and fails on the LAN IP (teeth verified by substitution).
The design we sketched (trunk + KV on the 3090, MoE experts on the iGPU) is not
expressible in mainline llama.cpp. Per-tensor cross-backend placement aborts in
the scheduler -- "pre-allocated tensor (blk.0.ffn_down_exps.weight) in a buffer
(Vulkan1) that cannot run the operation (NONE)" -- and both per-layer
cross-backend forms fail at load. CUDA + Vulkan cannot serve one model; the CPU
is the only fallback side that works, which is exactly why -ot exps=CPU runs.
What the same build does instead is the surprise: iGPU-only (Vulkan/RADV, PLE in
page cache) gives 32.64 tok/s shallow, 25.77 tok/s at 64k depth and 456.7 tok/s
prefill, with the 3090 idle at 0 %. That matches halogen's deployed ~25 tok/s --
but commits 116 of the host's 121 GiB against halogen's 95.4 held with 14.7
spare, so it is not deployable beside the co-tenants. Throughput parity, memory
bankruptcy: the fleet's two-endpoint split stands, now on evidence rather than
preference.
New page LLM/x4-pair-experiment.md carries the tensor inventory, the results
table, the two scheduler limits, reproduction commands, and the traps found on
the way. Both index pages also stop calling halogen retired -- the ik swap's
stale claims were still sitting in them.
The fleet now runs draft-dflash,ngram-mod, so the rule needs to be written down
where the next agent will look: draftless types (ngram family) are added to the
spec configuration unconditionally, while draft-mtp/dflash/dspark/eagle3 all
require a draft CONTEXT -- therefore one context-backed drafter plus any number
of ngram types is legal, and two context-backed ones never are ('draft-mtp,draft-dflash'
dies at load with 'model doesn't contain MTP layers'). Plus the operational
trap: ngram-mod prints no startup banner, so a working chain looks unengaged in
the journal.
The port gate was written for the cutover, when the vLLM proxy held 7770. Once
frank-medium itself owns the port, a re-deploy waits 90s for a socket that only
its own predecessor is holding and fails: 'FAIL: port 7770 is still held after
90s', with the old llama-server still serving scrapes. Stop the unit first, then
sweep a stray manual llama-server (the shape a benchmark harness leaves behind);
pkill -x, never -f, so the deploying shell cannot match itself.
ngram-mod is draftless (added to the spec config vector unconditionally, while
every model-backed drafter needs ctx_dft), so unlike MTP it can share the
process with DFlash2. Chained at the fleet's 48/64 window, over three shapes:
prose 1k/512 dflash 43.77 | ngram 17.72 (0 drafts) | chain 43.65
echo 1.5k/400 dflash 76.82 | ngram 102.50 | chain 200.58
structured/400 dflash 91.08 | ngram 17.80 (0 drafts) | chain 91.22
identity dflash 70.37 | ngram 17.77 | chain 70.40
On prose the chain is byte-identical to the drafter alone (same 119 drafts/49
accepts) -- that is what makes it a safe default -- and on prompt-echo traffic
it is 2.6x. VRAM is unchanged: the matcher is a 64 MiB host-RAM hash table.
The window sweep says keep the long one: 48/64 beats 16/32 (161.0) and 8/16
(113.8) on echo, and shortening it costs prose (43.9 -> 36.0) because the
matcher then proposes on ordinary fragments and the rejections still have to be
verified. That is the 'ngram hurts' result this fleet hit twice before.
Two documentation traps recorded with it: ngram-mod prints NO startup banner in
this build (its drafts are only visible behaviourally), and the harness's
'structured' shape is a negative control because its lines differ.
Two real defects from patching a rewritten file by line number: the CUT that
removed a duplicate parse_configs also took write_env with it (NameError after
the first configuration, with the unit already stopped), and the restore step
lived after the loop, so a crash left frank-medium DOWN. The endpoint is the
reason the harness exists; it now restores itself in a finally and shouts if the
service does not come back healthy.
ngram-mod's match window (n_min/n_max) decides which traffic it can see at all
-- the fleet's 48/64 preset is tuned for long verbatim echoes and finds nothing
in a repetitive-but-short output. Config entries now take |KEY=VALUE overrides,
so the window can be swept in the same run as the configurations it applies to.
The first run asked for a verbatim repeat with thinking ON, so the model copied
the passage inside reasoning_content, content came back empty for all three
configurations, and the cross-configuration identity check compared '' to '' --
a pass that proves nothing. echo and structured now disable thinking via
chat_template_kwargs, the echo row carries a copy_ratio (share of answer lines
that really are passage lines), and an empty answer in any shape is a loud
warning rather than a silent pass.
'structured' is new and answers the case the fleet actually runs on: repetitive
OUTPUT with no prompt overlap (record lists, tables, config rewrites), which is
where an n-gram matcher should pay and an echo-shaped test cannot tell.
A relative serve-script path plus a cd into its own directory double-counts the
prefix, so every configuration failed to start ('scripts/serve-frank-medium.sh:
No such file or directory'). The safety net did its job -- the unit was restored
-- but the run measured nothing.
Unlike MTP, ngram-mod needs no draft context: common/speculative.cpp adds the
ngram types unconditionally while the model-backed drafters are gated on
ctx_dft != nullptr, so 'draft-dflash,ngram-mod' is structurally allowed (and the
config vector's fixed order -- ngram first, then the drafters -- makes CLI order
irrelevant).
The harness grows the two things this question needs: --configs, and an ECHO
shape. ngram-mod can only ever match text that is present in the prompt, so
measuring it on prose calls it useless and measuring DFlash on echo calls it
wasteful; both shapes now run, and the greedy text of each is compared across
configurations because speculation may not change answers.
Four configurations of the same binary, identical prompts at temperature 0, the
unit stopped between runs, on the deployed payload:
none 17.7 tok/s card1 26.07 GB
draft-mtp n=3 32.4 tok/s 58 % accepted card1 29.35 GB
draft-mtp n=7 26.7 tok/s 33 % accepted card1 29.65 GB
draft-dflash 7 43.5 tok/s 41 % accepted card1 29.42 GB
DFlash2 is worth 2.46x over autoregressive and 34 % over the best MTP, and all
four produced IDENTICAL greedy text, so the 'lossless verification' claim in the
runbook is now a check rather than an assumption. MTP is not the cheaper option
either: its head loads ~3.3 GB against the drafter's 2.06 GB file, and deeper
drafting loses (n=7 accepts a third of its proposals, n=3 more than half).
The two CANNOT be chained, and the failure is structural rather than tunable:
'spec-type draft-mtp,draft-dflash' dies at load with 'context type MTP requested
but model doesn't contain MTP layers', because a chain has ONE draft context and
MTP builds against whatever model that context holds -- here the DFlash2 file.
The serve script therefore passes --spec-draft-model in the dflash branch ONLY,
and the seam now asserts exactly that (the wiring whose violation is this
error), plus that the deployment stays on the measured winner.
scripts/bench-spec-ab.py makes the comparison reproducible: the benchmarking
policy wants the runtime's own harness for this, and the numbers above were
otherwise unreproducible one-offs.
The loader prints no per-device ledger at this verbosity, so the claim is
measured rather than read: +131072 tokens of context added 4.83 GB to one card
and 4.84 GB to the other (both carry their own attention layers' KV), and
neither card's total usage could hold the 31.46 GB file. Worth writing down
because the retired shape WAS two full copies — that is what a replica is, and
it is why the 29.8 GB FP8 checkpoint could never run there.
Frank