Case Study: San Diego Supercomputer Center achieves balanced, massive-scale HPC serving 50,000+ users with Dell Technologies

How they designed Expanse to accommodate a vast array of workloads for both research and industry with on premises and cloud resources

The San Diego Supercomputer Center (SDSC) needed to deliver balanced, high-performance computing to tens of thousands of academic and industry researchers who run highly mixed workloads—from single‑core high‑throughput jobs and science gateways to GPU‑accelerated AI and large-scale simulations. Working with Dell Technologies, SDSC sought a future‑proof, cost‑efficient system that met NSF constraints and the center’s operational and utilization targets.

Dell Technologies delivered the Expanse supercomputer using Dell EMC PowerEdge servers with AMD EPYC processors, NVIDIA V100 GPUs and Mellanox HDR InfiniBand, producing a heterogeneous system with 5 petaflops peak performance (≈728 compute nodes, 52 GPU nodes/208 GPUs, ~93,184 cores), 12 PB Lustre and 7 PB object storage. The deployment doubled expected throughput versus the prior system Comet, is projected to serve more than 50,000 users, and enables SDSC to support a wide diversity of workloads while improving utilization, scalability and scientific discovery.

San Diego Supercomputer Center

Shawn Strande

Deputy Director