High Productivity Technical Computing

The HPTC facility has partnerships with Sun Microsystems and other computer and IT companies to provide and maintain high productivity computer systems, both hardware and software.

CEES is about productive Earth science and computer science research and education, not the fastest computers. To this end, the HPTC facility provides staff (professional and students) to support computer operations and research users.

HPTC hardware. High productivity hardware is the core of the facility. These systems deliver sufficient computational power and storage capacity to enable users to make quantum leaps in their computational research. Ideally, they should be easy to use as well. Several research groups in the school already have the necessary experience to benefit from large-scale parallel computing. In order to expand the local community of serious computational researchers, the center must facilitate the transition to a large-scale computing culture for those not yet skilled in developing numerical algorithms (often parallel). Our center will help clarify best approaches and best practices for various classes of problems. Our IT–computer partners play an important role in implementing a user friendly environment through which hardware and software are efficiently utilized and optimized for Earth science problems.

The CEES Grid hardware is organized into three resource clusters:

1. The Opteron Cluster is composed of 64 Sun v20z dual cpu nodes running Linux. Half of the nodes have 2GB of memory, and the other half have 8GB of memory. The head node is a Sun v40z with 4 dual core Opteron cpus and 16GB of memory. This cluster is designed for distributed programming and runs MPICH. The cluster software is Sun Grid Engine, and the network is Infiniband.
2. The Sparc Cluster is composed of 2 Sun servers running Solaris 10. The head node is a Sun 490 with 4 dual core Sparc cpus and 16 GB of memory, and the compute node is a Sun 6900 with 24 dual core Sparc cpus and 192 GB of memory. This cluster is designed for SMP programs and runs OpenMP. The cluster software is Sun Grid Engine, and the network is 10Gb ethernet.
3. The Tool Cluster is composed of 2 Sun v40z machines running Linux, each with 4 dual core Opteron cpus and 32 GB of memory. This cluster runs standalone programs and applications such as Matlab.

HPTC staff. To enable faculty and students to become effective users of the computational facilities, the center provides a small staff to assist users. We plan to support the following two staff positions:

• System Manager (responsible for traditional system administration tasks such as hardware and software installation, maintenance, and upgrades; user accounts and help; backups; documentation; and system monitoring and security).
• Numerical Analyst (responsible for assisting users with numerical and parallel algorithms and general technical support)
  In addition, we will employ students to provide additional hardware and software support.

CEES Home  |  Users  |  People  |  Research  |  Education  |  HPTC Facilities  
RAs & Fellowships  |  News & Publications  |  Map & Directions  |  Contact Us