DSAI Cluster

The Advanced Research Computing at Hopkins (ARCH) manages the High Performance Computing core facility at Johns Hopkins University. DSAI (our current AI system) consists of:

• 80 CPU nodes (8,960 cores)

• 15 nodes with 8x NVIDIA A100 (80GB) GPUs

• 16 nodes with 4x NVIDIA H100 (80GB) GPUs

• 16 nodes with 4x NVIDIA H100-NVL (96GB) GPUs

• 8 nodes with 8x NVIDIA L40S GPUs

• 5 PBs of WEKA storage

These are “shared resources”, therefore we expect all users to comply with policies on passwords, security guidelines, resource utilization and best practices, and privacy issues

If users have any questions about these policies, please send an email to help@arch.jhu.edu

• DSAI Quick Start
• Creating and Managing Allocations and Accounts
• Resources
• Navigating
• Slurm on DSAI