AI NODES
4 / 4 online
↑ 100% uptime
BikerLink · AI infrastructure
AI control room LIVE
Uno sguardo operativo su modelli, GPU e compiti in corso.
VRAM IN USE
49.0 / 84 GB
50.6% allocated
AVG GPU LOAD
52%
↑ 8.4% vs last hour
THERMAL HEADROOM
11°C
Peak 67°C · Horus
GPU UTILIZATION
Load across all nodes
Live load 52%Baseline 48%Updated 19:22:08
100%75%50%25%0%
−60m−45m−30m−15mnow
CURRENT QUEUE
Tasks & ownership
Telemetry summaryHorus · 72% complete
nowRoute analysis / T07Quebracho · BikerLink
2mContext warm-upNadir · BikerCuddle
4mEmbedding indexBowie · scheduled
idleINFERENCE LAYER
AI nodes & loaded models
SELECTED NODE
Horus busy
H
Llama 3.1 70B · IQ3_MThinkCentre TC · Linux · 2× RTX 3090
82%GPU load
VRAM ALLOCATION23.860 MBGPU 0 51% · GPU 1 49%
TEMPERATURE67°Cwatch threshold
ASSIGNED TASKTelemetry summaryby Quebracho Box
NETWORK PORT:8120healthy · 218 ms
CONNECTIVITY
6 / 12 usedPorts & links
8110Quebracho APIHTTP · BikerLink
8120Horus inferencegRPC · Quebracho Box
8130Nadir chatWebSocket · BikerCuddle
8140Bowie embeddingsHTTP · internal
AUDIT TRAIL
Activity log recording
TIMEEVENTDETAILORIGIN
19:22:08Telemetry heartbeatAll 4 AI nodes answered · 218 msbridge / local
19:21:42Task assignedHorus ← Quebracho Box · telemetry summarybridge / local
19:20:15Model loadedNadir · Mistral Small 24B · 8.6 GB VRAMbridge / local
19:18:56GPU split changedHorus · GPU 0 51% / GPU 1 49%bridge / local
19:17:11Port check8110, 8120, 8130, 8140 · healthybridge / local