FLEET//CONTROL — Local AI Server Fleet

Σ

Aggregate fleet throughput

0tok/s

▲ 0.0% peak today — · concurrency 0

Inference per GPU

—

pooled VRAM —

GPU UTILISATION GRID · nominal · thermal throttled · idle

Servers

0

—HEALTHY

Active GPUs

0/28

—AVG —%

Running models

0

—

Queued jobs

0

—

Current power draw

0 kW

of 16.0 kW budget · —%

PUE 1.09 · —/day

Tokens per second

per server

Active alerts

—

Power draw

watts · 15m buckets

VRAM envelope

GB stacked

Activity log

tail -f fleet.log

Fleet vitals

rolling 60s

Storage volumes

Thermal map

each bar = one GPU · height = utilisation · colour = die temperature

Job completion time

Queue depth

jobs waiting

p50 wait

—

p95 wait

—

throughput

—

Active

0

Waiting

0

Completed

0

Failed

0