Atlas Container Platform


Cluster Overview

Basic Information

Cluster Type

Kubernetes

Status

Healthy

Total Racks

4

Location

Datacenter

EU-CENTRAL-1

Halls

Hall C

Description

Kubernetes cluster for containerized workloads

Cluster Utilization

5.2%

88.9%

607 active jobs

Power Consumption

30.4 MW

PUE: 2 • 474.3 kW/node

GPU Health

N/A

All GPUs operational

Compute Performance

606.7K TFLOPS

303 jobs queued

Cluster Specifications

Compute Resources

Total Nodes

64

CPU Cores

4,096

Memory

32 TB

Storage

1 PB

GPU Configuration

Total GPUs

6070

GPU Models

MI250X, MI250X, MI250X, MI250X

Topology

SUPERPOD

Interconnect

ROCE_V2

GPU Utilization

89%

Network Configuration

Compute Fabric

ROCE_V2

Topology

TORUS

Bandwidth

61 Tbps

Latency

3053 ns

Management Subnet

10.163.255.0/24

Cluster Utilization

Loading cluster utilization data...

Rack Composition

Rack R1-1

STORAGE

Power

11.5 / 35 kW

Cooling

rear door

Temps

10°C → 50°C

Space

16/48U (32U free)

Rack R1-2

STORAGE

Power

12.3 / 35 kW

Cooling

rear door

Temps

26°C → 22°C

Space

16/48U (32U free)

Rack R1-3

STORAGE

Power

13.1 / 35 kW

Cooling

rear door

Temps

22°C → 30°C

Space

16/48U (32U free)

Rack R1-4

STORAGE

Power

13.9 / 35 kW

Cooling

rear door

Temps

18°C → 37°C

Space

16/48U (32U free)

Workload Scheduler

Type

KUBERNETES

Endpoint

https://k8s-master.example.com:6443

Version

1.28.2

Jobs Running

607

Jobs Queued

303

Configuration

Auto Scaling

Disabled

Power Capping

Disabled

Maintenance Window

Thu 14:00 (5h)

Metadata

Created

7/5/2025, 2:20:14 PM

Last Updated

7/5/2025, 2:20:14 PM

Tags

high-priority,gpu-optimizedhigh-priority,gpu-optimizedhigh-priority,gpu-optimizedhigh-priority,gpu-optimized