Shaheen (supercomputer)

Shaheen is a supercomputer owned and operated by King Abdullah University of Science and Technology, Thuwal, Saudi Arabia. It consists primarily of a 36-cabinet Cray XC40 supercomputer. Built in partnership with Cray, Shaheen is intended to enable KAUST Faculty and Partners to research both large and small-scale projects, from inception to realization. Shaheen's creator is Majid Alghaslan, who was then KAUST's interim chief information officer. Alghaslan also named the machine.[1]

Shaheen, named after the Peregrine Falcon, is the largest and most powerful supercomputer in the Middle East with a processing power of 5.54 petaflops with 196,608 cores. The first generation of Shaheen (2009-2015) was an IBM Blue Gene/P, originally built at IBM's Thomas J. Watson Research Center in Yorktown Heights, New York, and moved to KAUST in mid-2009. The second generation of Shaheen, project name "Shaheen II", was brought into service in the summer of 2015.

Systems

Shaheen II

The second generation of Shaheen has the following specifications:[2]

  • Cray XC40 supercomputer with DataWarp technology
  • Cray Sonexion 2000 storage system

Shaheen I

The first generation of Shaheen included the following functional elements:

  • 16 racks of Blue Gene/P, having a peak performance of 222 Teraflops
  • 164 IBM IBM System x 3550 Xeon nodes, having a peak performance of 12 Teraflops

Performance

Shaheen I's performance and computing capabilities included:[3]

  • 65,536 independent processing cores.
  • 10 Gbit/s access to world's academic and research networks.

The file system and Spectra Logic TFinity tape system (https://spectralogic.com/2015/07/13/spectra-logic-tape-provides-archive-and-backup-solution-to-kaust-supercomputer/) were mounted across both the Blue Gene system and the Linux cluster. All elements of the system were connected together on a common network backbone that is accessible from all campus buildings. The systems were also be accessible from the Internet.

Services

The Shaheen system at KAUST Supercomputing Laboratory (KSL) is available to help KAUST users and projects, to provide training and advice, to develop and deploy applications, to provide consultation on best practices and to provide collaboration support as needed.

KAUST Faculty will have access to:

  • General support for Shaheen facility use, including usage scheduling of Shaheen and peripheral systems
  • High-performance computing support for "Grand Challenges" by collaboration with the Center to deliver fundamental breakthroughs in specific areas of research
  • Collaboration to provide high-performance computing applications, middleware, library, algorithm support and enablement services
  • Applications Enablement where users can task the CDCR to develop, enable, port and scale key applications
  • High-performance Computing Program Best Practice Management techniques
  • Participation with KAUST researchers in external projects
  • Training on high-performance computing systems management, programming, applications tuning and algorithms

Research Supported

KAUST, using the Shaheen systems, will focus on four specific research thrusts:

Data sets for this research will be unique in that they will come from the Saudi Arabia region, focusing on areas such as oil and gas reserves, Red Sea data, and other areas distinctive to KAUST.

Restrictions

Although KAUST does not support any kind of discrimination against any students or faculty based on religion, sex, or national origin, IBM and Cray have to comply with US Export regulations governing exporting high-end computing technology. As per the regulations[4] a restriction had to be made to deny nationals of Syria, Iran, Sudan, Cuba, and North Korea from access to the supercomputer.

Institutional Partners

The Shaheen system at KAUST is made possible through a joint collaboration between the Cray and KAUST. In addition to IBM and Cray, KSL has partnered with the following partner research institutions and organizations:

See also

References

This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.