2026-07-26 - 2026-08-02
Overview
3 Issues created by 1 user
Opened
#11 comfyui's vram_budget_mib is 11000, measured usage is 4346 — the broker refuses image jobs that would fit comfortably
Opened
#12 comfyui's VRAM need is not a constant: it varies cold-vs-warm (8410 vs 6170 MiB) and with the requested resolution
Opened
#13 The broker can reclaim VRAM it did not allocate, but nothing stops a direct client taking it — a bare ollama run still OOM-kills whisper
10 Unresolved Conversations
Open
#3
mGPUmanager: ollama not evictable + no game-awareness + OLLAMA_KEEP_ALIVE=24h starve the game/FLUX
Open
#10
STT glossary drift: consumers.yaml stt_prompt is hand-maintained and diverges from mVoice lexicon.yaml (missing mBrian → 'mBryan')
Open
#6
Broker must be the SOLE VRAM authority — close bypass gaps so out-of-band ollama holds get evicted (imagen starved by 24h keep-alive)
Open
#7
STT big-model via GPU manager: mVoice eager-loads at startup + PWA bypasses the broker → big model OOM-crashes on a full GPU
Open
#9
STT forces German on every utterance: -l de hardcoded + language never forwarded, so English voice is translated to German
Open
#4
Broker reports per-consumer gpu_resident_mib=0 for externally-started consumers → eviction finds "no candidates" and never reclaims their VRAM
Open
#2
Add a generic GPU-lease primitive (acquire/renew/release) for long-running async consumers (ComfyUI/FLUX)
Open
#5
Scheduler must evict ollama for higher-priority image leases (vram_managed guard defeats the manager's purpose)
Open
#1
Phase 1: GPU-Manager — Schritte 0–5 (ComfyUI systemd → mVoice load/unload → Broker MVP → wa.sh migrate → Queue → LRU-Eviction)
Open
#8
unload(): whisper-server systemd branch marks consumer unloaded without stopping the unit (latent VRAM-not-freed bug)