GPU online
18,432
99.982% fleet healthy
GPU cluster fabric
Live capacity by training, inference and reservation lanes.
Node rack map
Click a node to inspect GPU, thermal and workload placement.
MEL-A2-N017
Healthy
Workload
llama-ops-ft-92
GPU temp
64.8°C
Node power
8.7kW
Network
342Gbps
GPU utilization
Training, inference and reserved lanes.
Chart library unavailable. Telemetry tables continue to update.
Queue wait
Placement latency by priority class.
Chart library unavailable. Telemetry tables continue to update.
Energy and cooling
IT load, PUE and thermal headroom.
Chart library unavailable. Telemetry tables continue to update.
Fabric throughput
East-west and storage traffic.
Chart library unavailable. Telemetry tables continue to update.
Job queue
Scheduler view for high-value training and inference workloads.
| Job | Tenant | Lane | GPUs | ETA | State |
|---|
Tenants and SLA
Quota burn, reserved capacity and service-level exposure.
Thermal plant
Cold aisle sensor band by zone.
Cooling load
6.8MW
Headroom
18%
Site topology
Primary region and edge compute availability.
Runbook timeline
Upcoming maintenance and automated actions.
Event log
Scheduler, facility and security events.