Cutting-Edge Computational Technology

Driven by the exponential growth of computational power, scientific computing is now radically transforming our research philosophy by enabling the simulation of many complex flow phenomena across a broad range of scales in natural and engineered systems with an unprecedented degree of realism. Coupled with our state-of-art measurement techniques and unique experimental facilities, our simulation-based expertise has uniquely positioned the laboratory to make far-reaching advances in the major societal problems of our time in energy, the environment and human health. 

SAFL has two High Performance Computing (HPC) Beowulf-style computer clusters with execution and compute nodes connected by low-latency/high-throughput local interconnects (InfiniBand), including:

  • Aegean: a 2,496 core 2.6 GHz AMD Interlagos system with 2GB RAM/core 
  • Paros: a 1,092 core 3.06 GHz Intel X5675 cluster with 4GB RAM/core

We also have 240 cores of Paros-class interactive nodes provisioned via Openstack for interactive use. The storage associated with the clusters is a 262TB Lustre-based storage system with distributed metadata servers (MDS) and object storage servers (OSS), all connected via the IB network in order to handle high I/O needs of the programs. A secondary 400TB storage system based on Serial Attached SCSI (SAS-2) disks with redundant controllers serves the long-term storage and day-to-day needs of the cluster

Examples of projects using our high performance computing includes ongoing simulation-based research aimed at understanding how sediments are transported in waterways; restoring deltas, rivers, and streams; optimizing wind and hydrokinetic turbine layouts; and improving the design of life-saving medical devices.

 

Contact safl@umn.edu to learn more about our computational team and facilities.