Specifications
- H200 nodes (4-way)4× NVIDIA H200 NVL (141 GB HBM3e each) · 64 CPUs · ~750 GB RAM
- H200 nodes (8-way NVL)8× NVIDIA H200 NVL · 64 CPUs · ~1,500 GB RAM
- H200 nodes (8-way SXM)8× NVIDIA H200 SXM · 192 CPUs · ~2,200 GB RAM
- B40 nodes (4-way)4× NVIDIA RTX PRO 6000 Blackwell (96 GB GDDR7 each) · 64 CPUs · ~512 GB RAM
- Schedulingmulti-tenant SLURM · up to 8 GPUs/node
- Queuesh200x4 · h200x8 · b40x4 (+ -long & debug variants)
- AccessSSH & Open OnDemand
Best suited for
- AI & machine learning
- Biomedical imaging
- Molecular modeling & NLP
- Scientific computing & data science
Getting access
Request access through the HPC support ticket system. Debug queues with higher priority and lower resource limits are available for quick testing before production runs.
New to HPC? Follow the get-started path — account, DUO, first login, first job — then come back for cluster-specific queues.
Other clusters
