Home \ Facility

Facility

A number of problems of interest to theoretical researchers lie at the forefront of structure formation and evolution of the universe. The fundamental problems are computationally challenging, intrinsically three dimensional and characterized by a wide range of length scales and time scales. In order to attack such problems requiring extensive memory and/or evolutionary time requirements, TIARA provides access to computational resources using many processors on clusters, visualization servers, and software, as outlined below.

Theoretical Clusters

XL Cluster (89 nodes, 1,456 cores, 6.3TB memory)
85 nodes	E5-2640 v3 @ 2.60GHz x 2	64GB RAM
4 nodes	E5-2680 v3 @ 2.50GHz x 2	256GB RAM
PX Cluster (27 nodes, 432 cores, 1.7TB memory)
27 nodes	E5-2690 @ 2.90GHz x 2	64GB RAM
TC Cluster (50 nodes, 600 cores, 2.3TB memory)
22 nodes	X5660 @ 2.80GHz x 2	48GB RAM
28 nodes	X5670 @ 2.93GHz x 2	48GB RAM
OC Cluster (32 nodes, 256 cores, 1.1TB memory)
32 nodes	X5560 @ 2.80GHz x 2	32GB RAM
GPU Cluster
1 nodes	E5-2620 v4 @ 2.10GHz x 2 NVIDIA Tesla P100-SXM2 (3,584 CUDA Cores, 16GB) x 2	256GB RAM
3 nodes	E5-2603 v2 x 2 NVIDIA TITAN X (Pascal) (3,584 CUDA Cores, 12GB) x 2	32GB RAM
Visualization Servers
2 nodes	X5675 @ 3.07GHz x 2 NVIDIA Quadro 4000 (256 CUDA Cores, 2GB) x 2	96GB RAM
1 nodes	X5450 @ 3.00GHz x 2 NVIDIA Quadro FX 4600 (128 CUDA Cores, 768MB)	32GB ram
Software
IRAF, LAM-MPI, MPICH, OpenMPI, OpenMP, Intel C/Fortran Compiler, PGI, FFTW, HDF, SuperMongo, Zeus3D, ZeusMP, pgplot, Visit, MatLab, IDL

Cluster Report

A monitoring system has been implemented to allow users to remotely view live or historical statistics for all nodes, such as CPU load averages or network utilization.