Advanced Computation & Visualization
The Nuclear Engineering Division has three HPC clusters as its advanced computational Multiple-Instruction Multiple-Data platform for performing computations of engineering mechanics, fluid dynamics, Monte Carlo simulations and engineering analysis. The three clusters are comprised of one 15 dual core node Ethernet, a 46 multicore (368 cores) Infiniband, as well as a 8 node multicore/large memory (256 cores/128GB per node) Infiniband cluster.
The front-end servers have both Intel Xeon and Opteron processors and one front end node is a large
memory system with 198 GB of RAM for CFD Meshing. The compute nodes have a variation by cluster of
both Intel Xeon processors running as well as Opteron. NE also has three large memory non-clustered machines for large meshing at 512GB and 1TB of total memory.
The two multicore clusters have their file systems provided by a new high performance fileserver from PANASAS (PAS 8).
The operating system's primarily used on the cluster are CentOS.
PGI Workstation 3.1 (F77, F90, HPF, C++, C), Intel Fortran Pro, Absoft F90 Pro, MPI.
The following applications have
been ported to the cluster:
- Monte Carlo,
- Abaqus and
Part of the two HPC clusters Eddy and Eddy2 is shown in the figure above.
Last Modified: Fri, May 30, 2014 1:48 PM