Start: Wednesday, 11 July 2012 @ 06:00

End: Thursday, 12 July 2012 @ 16:00

Description:
The Cray Centre of Excellence for HECToR delivered this course on the Cray XE6 over two days. It introduced new and existing users to the more powerful and advanced features of the hardware and softwares stack available. The programme will included:

    A detailed overview of the “Interlagos” Opteron processor architecture and Gemini network interconnect.

            Details on using the Interlagos cores in fully populated or core-pair mode.


    Demonstrations of the additional flexibility and functionality available with the Cray MPI library:

            Methods for customising the placement of ranks and processes to improve individual application communication performance.

            Mechanisms for improving the overlap between communication and computation within an application.

            Using and optimising MPI-IO for efficient file system performance.


    Guidance on use of Cray supplied optimisations of the standard installed scientific libraries.

    Tutorials on Cray debugging and analysis tools, CrayPAT, ATP, STAT and Fast-track debugging.

    General tips and tricks from the Cray Centre of Excellence for HECToR on optimising applications on Cray architectures.

Attendees benefit from experience with or an interest in Cray hardware.

Pre-requisite Programming Languages:

Knowledge of Fortran, C or C++, and MPI would be beneficial. It is not possible to do the exercises in Java.

https://events.prace-ri.eu/event/70/

Event type:
  • Workshops and courses
Cray XE6 Performance Workshop @ EPCC https://tess.elixir-europe.org/events/cray-xe6-performance-workshop-epcc The Cray Centre of Excellence for HECToR delivered this course on the Cray XE6 over two days. It introduced new and existing users to the more powerful and advanced features of the hardware and softwares stack available. The programme will included: A detailed overview of the “Interlagos” Opteron processor architecture and Gemini network interconnect. Details on using the Interlagos cores in fully populated or core-pair mode. Demonstrations of the additional flexibility and functionality available with the Cray MPI library: Methods for customising the placement of ranks and processes to improve individual application communication performance. Mechanisms for improving the overlap between communication and computation within an application. Using and optimising MPI-IO for efficient file system performance. Guidance on use of Cray supplied optimisations of the standard installed scientific libraries. Tutorials on Cray debugging and analysis tools, CrayPAT, ATP, STAT and Fast-track debugging. General tips and tricks from the Cray Centre of Excellence for HECToR on optimising applications on Cray architectures. Attendees benefit from experience with or an interest in Cray hardware. Pre-requisite Programming Languages: Knowledge of Fortran, C or C++, and MPI would be beneficial. It is not possible to do the exercises in Java. https://events.prace-ri.eu/event/70/ 2012-07-11 06:00:00 UTC 2012-07-12 16:00:00 UTC [] [] [] workshops_and_courses [] []