Coming Soon!

How to get access

Access model coming soon! Fill in the expression of interest form.

Contact sih.info@sydney.edu.au with any questions.

Inference end points

Coming soon!

Raw GPU Access

http://Run.ai

Machine Specs

Operating System

• NVIDIA Base Command Manager with DGX OS 7

  • Installed on head nodes, Kubernetes master nodes, and DGX nodes

  • Includes built-in LDAP authentication

Job Scheduler & Cluster Management:

• Kubernetes (K8s)

  • Three master nodes for managing the cluster

  • DGX nodes function as Kubernetes worker nodes

• Run:AI

  • Installed on Kubernetes for workload scheduling and AI workload orchestration

  • Includes namespaces, secrets, backend, and API server configurations

GPU nodes

3 x NVIDIA DGX H200 with:

• 8 x H200 per node.

• 1128 GB VRAM per node.

CPU nodes

5 x Dell PowerEdge R760 Server with:

• Primary and Secondary Head Nodes with High-Availability configuration for failover protection

• Three master nodes for managing Kubernetes

• 2x Intel Xeon Gold 5418Y 2GHz, 24Cores/48Threads

• 16x 32GB RDIMM, 5600MT/s, Dual Rank

• BOSS-N1 controller card + with 2 M.2 480GB (RAID 1)

• 6.4TB Enterprise NVMe Mixed Use AG Drive U.2 Gen4 with carrier

• Broadcom 57416 Dual Port 10GbE BASE-T Adapter, OCP NIC 3.0

• Mellanox ConnectX-6 DX Dual Port 100GbE QSFP56 Network Adapter

Storage

• DDN EXAScaler Parallel File System 1 PB

  • Provides high-performance shared storage to DGX nodes

  • Connected via 8x 200GbE Active Optical Cables

1 x Controller: DDN ES400NVX2-NDR200-SE with:

• 2 x SE2420-EBOD NVMe Expansion Enclosures with 24 NVMe drives each

• Total 48 x 30.72TB QLC NVMe G4 4K SSD drive

Networking

• Cumulus Linux

  • Runs on NVIDIA Spectrum-2 200GbE switches

  • Supports MLAG, VLANs, and routing configurations

• InfiniBand Networking for high-speed GPU-to-GPU communication

• Out-of-Band (OOB) Management