Cluster Information

Your cluster’s firewall appliance. It provide network protection and remote access services. All traffic into and out of your cluster is mediated by the FortiGate firewall appliance.

Endpoint: https://127.0.0.1:10443/
Username: rakuten
Password: provided by Lambda during hand-off

This portal provides a simple way of directly accessing your cluster from your browser.

Endpoint: https://10.148.0.200
Username: admin
Password: provided by Lambda during hand-off

This portal enables you to do various administrative tasks, including user management. Note that the admin web portal has two endpoints for high availability. If you want to do administrative tasks, first try using gi-primary; if it’s down, try gi-secondary.

Username: admn
Password: provided by Lambda during hand-off

• Software: Ubuntu 22.04
• Chassis: Supermicro A+ Server 1114CS-TNR
• CPU: AMD EPYC 7513 (32 cores, 2.6 GHz)
• Memory: 256 GB (8 x 32 GB 3200 MHz ECC RAM)
• OS storage: 1.92 TB SATA SSD
• Extra storage: 1.92 TB NVMe SSD
• NIC: ConnectX-6 VPI, dual-port, 200 gbps

• Software: Ubuntu 22.04, OFED, Lambda Stack (PyTorch, TensorFlow, NVIDIA drivers, CUDA, cuDNN)
• Chassis: AS-8125GS-TNHR
• CPU: AMD EPYC 9654 (96 cores, 2.4 GHz)
• GPUs: 8x NVIDIA H100 GPUs, SXM5, NVLink and NVSwitch connected
• Memory: 1.5 TB (24 x 64 GB 3200 MHz DDR4 ECC RAM)
• Storage: 1x 3.84 TB NVMe, 2x 7.68 TB NVMes
• NICs: 8x ConnectX-7, single-port, 400 gbps

• Server manual
• OFED manual