Available Hardware and Software

The CTP Linux Cluster currently contains one head node and 36 computing nodes. Thirty four computing nodes (node01-node34) have two Intel Xeon E5 processors interconnected with 10GbE network and 128 Gb RAM memory each. Computing nodes 35 and 36 are composed as four nodes each assembled in a single box.

Twenty eight computing nodes (node09-node36) are augmented by graphics processing units (GPUs) NVIDIA Tesla Kepler K20 that provides significant improvement of the computational capabilities. The GPU CUDA 6.0 compiler and libraries are installed on these nodes.

The following software is also installed on the Cluster:

  • Intel Software Development Suite
  • VASP 5.3.5
  • COMSOL Multiphysics with Wave Optics Module

The latest GCC compiler with Open MPI parallel libraries and latest Python distribution are available on the Cluster via the module system. For comprehensive documentation on the module system, follow this link: Module manual.

  • Matlab R2016a is installed on node 28. Please see an example of the PBS submission script in the directory /usr/local/share/torque/examples/matlab_job on the head node of the cluster.
  • Mathematica 10.4 is installed on node 27. An example of the submission script is in the folder /usr/local/share/torque/examples/mathematica_job on the head node of the cluster.

Access to the Cluster

The Cluster can be accessed through internet from inside and outside the College. The Cluster address is: ctp.citytech.cuny.edu.

  • From Linux and Mac OSX operating systems: Connection to the Cluster is available via ssh from an X-terminal on the user’s machine.
  • From Windows operating system: To access the Cluster from Windows system, users need to install putty terminal.

Torque (PBS) system

The CTP cluster is using Torque as the queuing system. To submit a job to the queue, you need to create a command file (a submission script). Basic examples of the scripts for serial and parellel jobs can be found on the head node of the cluster in the directory /usr/local/share/torque/examples/.

The job submission command is

  • qsub submit_job.pbs
where submit_job.pbs is the submission script.

To see the jobs submitted to the queue, use

  • qstat
To cancel a submitted job, use
  • qdel job-id
where job-id is the job id number shown by qstat command.

For the advanced usage, please see the following link: Torque reference guide.

File Transfer

Files can be transferred to and from the Cluster with standard scp or rsync commands from Linux and Mac OSX operating systems. On Windows system, users need to install one of freely available sftp clients (for example, WinSCP).

Contact Us

With any question about access or submitting jobs on the cluster, contact the CTP Computing Cluster Team:

Cluster Administrator:

Dr. German Kolmakov
Physics Department
Phone: 718-260-5329
Email: gkolmakov@citytech.cuny.edu

Software Administrator:

Dr. Ilya Grigorenko
Physics Department
Phone: 718-260-5279
Email: igrigorenko@citytech.cuny.edu

Acknowledgments to CTP

Please include the following acknowledgment on the materials resulting from the use of CTP resources:

“The author(s) are grateful to the Center for Theoretical Physics of New York City College of Technology for providing computational resources.”

Please, send a copy of the materials to the Cluster administrator.