Ollama limited to 4 cores and 2 GB, matching the planned CT-materio_embed
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ShDspHaZGwHoRZCsPBJzsM
This commit is contained in:
@@ -80,6 +80,12 @@ services:
|
|||||||
environment:
|
environment:
|
||||||
# garde le modèle chargé en mémoire (pas de rechargement à froid entre deux appels)
|
# garde le modèle chargé en mémoire (pas de rechargement à froid entre deux appels)
|
||||||
- OLLAMA_KEEP_ALIVE=-1
|
- OLLAMA_KEEP_ALIVE=-1
|
||||||
|
# Ressources du futur CT-materio_embed (E4) : 4 cœurs, 2 Go. Cœurs physiques 0-3 du
|
||||||
|
# laptop (lscpu -e=CPU,CORE) ; Ollama ignore cette limite pour ses threads, le module
|
||||||
|
# envoie options.num_thread (materio_ai.settings:num_thread = 4).
|
||||||
|
cpuset: "0-3"
|
||||||
|
mem_limit: 2g
|
||||||
|
memswap_limit: 2g
|
||||||
volumes:
|
volumes:
|
||||||
- ./var_ollama:/root/.ollama
|
- ./var_ollama:/root/.ollama
|
||||||
networks:
|
networks:
|
||||||
|
|||||||
Reference in New Issue
Block a user