ComfyUI's CLIP.encode_from_tokens() calls load_model() -> model_management.load_models_gpu()
on every encode. With the dynamic VRAM loader (comfy_aimdo, 0.4x+) that re-runs
ModelPatcherDynamic.load() for a model that is already resident: a walk over every module plus
free_memory bookkeeping, visible as one
Model MOSSImageTEModel_ prepared for dynamic VRAM loading. 7002MB Staged. ...
log line per prompt (N prompts -> N+1 lines). USCG's pre-encode paid it once per prompt, and
again on the single-encode paths (model switch, distributed, upscale, Florence2 hires).
Measured on the 5060 Ti (ComfyUI 0.42.5), MOSS-Image TE = Qwen3.5-9B w4a8, same process state:
template path (CLIPTextEncode), per prompt ....... 0.92 s (re-prepare every encode)
USCG batch encode with pinned_clip, per prompt ... 0.84 s (one prepare per pass)
USCG job, 21 prompts + 21 one-step images ........ 36.8 s -> 33.0 s
for scale: SDXL 0.15 s, Krea 2 0.29 s, Qwen-Image 0.32 s per prompt on the template path
The rest is the encoder forward itself: ~0.55 s fixed + ~2 ms/token for Qwen3.5-9B (24 of its
32 layers are Gated DeltaNet run through the pure-torch chunked recurrence, fp32 compute), so
a 60-prompt MOSS grid still spends ~50 s encoding before the first image. That part is
identical to the official workflow.
batch_encoding.pinned_clip(clip, texts): load once through load_model() (sized by the longest
text so any memory estimate still runs), shadow clip.load_model with a no-op for the block,
restore on exit or exception, nesting safe. encode_pair_cached(): positive+negative with one
load, nothing when both are cached. The pre-encode summary line now prints total time and
ms/prompt.
Applied to batch_encode_with_cache, the model-switch pre-encode in the main loop, the
distributed group pre-encode and per-job encode, the deferred-upscale re-encode fallback, the
dashboard upscale runner and the Florence2 hires step.
Tests: tests/test_pinned_clip.py (10) against the real batch_encoding module; the conftest
stub gained pinned_clip/encode_pair_cached. 137 passed, same 15 pre-existing torch-stub failures.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
MOSS-Image (Flux.2 latent: 128ch, 16 px per cell) jobs were built with the
SD-era `h // 8`, so every image sampled a latent with 4x the cells and decoded
at 2x the requested size (832x1216 -> 1664x2432): ~10x slower than the official
ComfyUI template (~120 s vs ~10 s per image on the 5060 Ti).
- model_loader: get_latent_downscale_ratio() reads latent_format.spacial_downscale_ratio
(8 for SD/SDXL/Flux.1, 16 for the Flux.2 family); make_empty_latent() builds the
job latent and warns when WxH is not a multiple of the ratio. Both orchestrator
latent sites (main loop + distributed worker) use it; hires-fix latent upscale,
tiled hires sampling and VAE tiling divisors no longer assume 8.
- CLIP type: resolve_clip_type() warns on unknown types instead of silently using
stable_diffusion, and when clip_type is left at the default it follows the
detected model family (MOSSImage -> moss_image, Flux2 -> flux2, ...). MOSS loaded
as a generic Qwen3.5 TE before (wrong chat template). The Builder's CLIP Type
list is now read from ComfyUI's CLIPLoader/DualCLIPLoader (static list = fallback,
now including moss_image); LS_CACHE_VERSION 3 refreshes cached lists.
- Skip Layer Guidance: new Extra Model & Sampling sub-group wrapping
SkipLayerGuidanceDiTSimple (double/single layers, start/end %), plumbed like Deep
Shrink (builder emission, expand_configs, orchestrator, generate_image). Defaults
match the official MOSS-Image workflow (single block 22, 0-50%).
- tests: latent geometry + clip type resolution, dynamic clip type lists, SLG
config emission/expansion (24 new; the 15 pre-existing torch-stub failures unchanged).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Third companion plugin in the extraction series. Completes the cleanup —
main USCG now has zero yara findings and should pass Comfy Manager
submissions without manual review.
Phase 2 renamed the routes to /uscg-distribution/* during the extraction
of the Distributed companion plugin. ProjectStructure.md (8 occurrences)
and README.md (6 occurrences) still referenced the old paths.
Caught alongside the worker URL bug during Phase C smoke test.
Re-add AI-CONTEXT.md, ProjectStructure.md, README.md, and Roadmap.md
to git tracking. Only docs/plans/ remains gitignored as local-only.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
ProjectStructure.md, README.md, Roadmap.md, AI-CONTEXT.md, and docs/
are local development references — not needed in the repo.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add distributed processing features (multi-worker, master text
encoding, model sync, per-worker logging)
- Add comprehensive AI development notes with architecture diagrams,
file dependency graphs, gotchas, and testing instructions
- Update file structure to include all 21 Python modules, distribution
files, and conf-builder-distribution.js
- Add distribution API routes reference
- Add distribution module descriptions and communication patterns
- Update data flow to include distribution pipeline
- Add 8 new critical development gotchas (13-20)
- Add new changelog entry for March 2026
- Bump version to 3.0
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
