CAMS/Math Computation Laboratory

Facilities, Equipment and Other Resources

Computing Equipment

High quality facilities supporting numerical computation are essential for the Department of Mathematical Sciences (DMS) and the Center for Applied Mathematics and Statistics (CAMS) at NJIT to fulfill their educational and research missions. Thus DMS and CAMS, with the help of SCREMS, CSUMS, UBM, and MRI grants from NSF, together with the generous support of NJIT, have maintained the CAMS Math Computation Laboratory (CMCL) for the research needs of their members since 1989.

Computational support provided by CMCL for the proposers consists of the workstations and desktop PC’s that are networked and available to investigators in their offices, plus other more major, shared facilities of the CMCL (see Table 1).

Table 1: CAMS/Math Computational Laboratory Facilities

Model Cores Processor & speed/GPU & max flops Storage / RAM
Intel multi-core 392 Intel Xeon, 2.2 to 2.53 GHz 9872 GB
Nvidia multi-GPU 15,320 NVIDIA Tesla K20(m), 1.17 Tflops 32 GB

The DMS has expanded its “Stheno” cluster in stages since its first server became operational in 2011. The cluster is intended to be used to test, debug, and run message-passing interface (MPI) codes. It now has 32 nodes and 392 cores, 3,840 GB of RAM, and 9,872 GB of local disk storage. Two servers of the cluster contain GPU’s, which now total 6, with a total of 32 GB of GPU RAM. The GPU’s are currently CUDA capable and are intended for general purpose computation on GPU-accelerated computing nodes.

The DMS also has its “Gorgon” cluster, which has been expanded sequentially since it became operational in 2010. This cluster is intended for jobs that require large memory, and for parallel computations that use the OpenMP application programming interface. It is now a 32 core system, with AMD Opteron 6134 processors running at 2.3 GHz, and a total of 64 GB of shared memory.

All computational facilities are maintained by the Academic and Research Computing Systems (ARCS) group, headed by its director, David Perel.

Recognizing the need to support the scientific and engineering computing that is essential to research efforts across the campus, NJIT provides all faculty, postdocs, and graduate students access to centralized computing servers for research purposes. These recently received a significant upgrade as part of a substantial donation by Linode, which is a Linux-based cloud hosting company based in New Jersey. The NJIT cluster, “Kong”, now has a total of 359 nodes, 2,812 CPUs, 22,704 GB of RAM, and a disk storage of 318,800 GB. Processors are all AMD Opteron or Intel Xeon models, with speeds from 2.2 GHz to 2.8 GHz. It also features two 2-GPU nodes (NVIDIA Tesla K20x). Each GPU has 2,688 cores, with a peak performance double precision rate of 1.31 Tflops, and 6 GB of RAM.

The Office

The DMS assigns an individual office to the faculty, postdoctoral associates, and common offices with computers and other equipment to graduate students. In addition, a conference room and the CAMS Reading Room are available for formal and informal research meetings.

Other

The DMS is the base of the Center for Applied Mathematics and Statistics (CAMS) to which all Investigators belong. CAMS supports research in the mathematical sciences at NJIT by preparing a CAMS Annual Report, series of CAMS Technical reports (in electronic form, available for public use at the CAMS website), maintaining a weekly colloquium on Applied Mathematics and Statistics, a weekly seminar in biomathematics, and biweekly seminars in fluid dynamics, wave propagation, and applied statistics. DMS and CAMS also sponsor a major conference on “Frontiers in Applied and Computational Mathematics,” held annually at NJIT.

Software

The software that is available on the UNIX workstations of the CAMS laboratories includes along with MATLAB signal processing, neural network, wavelet, and symbolic math toolboxes.

  • Compilers: Fortran77, Fortran90, C, C++, Ada, Pascal
  • Numerical processing packages and libraries: Matlab, Mathematica
  • Symbolic processing packages
  • Graphics packages and libraries: Tecplot, gnuplot, Showcase, xfig, xmgr
  • Statistical computing: S-PLUS, SAS, IMSL
  • Text processing: Latex, nedit