Opus High-Performance Computing Cluster


opus_art_bg_800x803The Opus High-Performance Computer (HPC) provides the UNF Taylor Engineering Research Institute with access to massively parallel computing capabilities for ocean and atmospheric modeling.  Opus was purchased in 2012 and custom built by Penguin Computing.


Opus is a Beowulf-type cluster with 16 compute nodes.  Each node consists of dual, 12-core AMD Opteron 6234 2.4 GHz CPUs, 64 GB RAM, and 450 GB local scratch space.  Eight of these nodes include nVidia Tesla M2070 6 GB GPUs, significantly enhancing the capabilities of the HPC on some types of computation.  In addition, Opus has access to about 50 terabytes of high-performance storage.


The operating system, like over 95% of high-performance computing systems, is GNU/Linux (specifically, CentOS).  The primary application hosted by Opus is ADCIRC, a system of computer programs for solving time-dependent, free surface circulation and transport problems in two and three dimensions.  Additional software applications can be compiled to run on Opus, utilizing a variety of implementations of message-passing interface and compiler suites.


Early LINPACK benchmarks rate Opus at 3.7 TFLOPS (CPUs only) and up to 7.8 TFLOPS with GPUs.

 

The Opus HPC is managed by UNF's Research Technology Services unit in the College of Computing, Engineering & Construction.