Systems

Research Computing Platforms

The current research computing cyberinfrastructure at the University of Louisville includes: a general-purpose, high-performance distributed-memory cluster, five high-memory SMP systems, and an informatics cluster. All of the systems are housed in the University's secure data center and are administered by a team of specialized HPC system administrators and supported by a team of research computing consultants with experience in HPC software and database design, development and optimization.

All research systems share over 400 TB of high-performance (GPFS-based) storage.

Facts about the distributed-memory cluster:

• 530 IBM iDatplex nodes each with two Intel Xeon quad or hex-core processors for 5112 total cores

• Each node has between 16 GB and 48 GB memory

• Each core has 2 GB to 4 GB of memory

• 14 nodes are equipped with dual NVIDIA M2050 or M2070 Tesla cards

• Estimated peak performance is over 50 TFLOPS

Facts about the SMP systems:

IBM P570

• 16 IBM Power6 4.7GHz CPUs

• 128 GB memory

Four high-memory SMP systems each with

• Four 8-core Intel Xeon 3.3GHz CPUs (32 cores per system)

• 768 GB memory per system

• 8 TB High-performance local scratch storage per system

The four-node, dedicated Spark/Hadoop informatics cluster has 50 TB of high-performance storage for high-reliability big data applications.

Contact us at it-hpc@louisville.edu with any questions or for additional information.