SciNet News May 2012

May 5, 2012 in for_researchers, for_users, newsletter

SYSTEM CHANGES

  • GPC: The GPC has been upgraded to a low-latency, high-bandwidth Infiniband network throughout the cluster. Several significant benefits over the old ethernet/infiniband mixed setup are expected, including:

    • better I/O performance for all jobs
    • better job performance for what used to be multi-node ethernet jobs (as they will now make use of Infiniband),
    • for users that were already using Infiniband, improved queue throughput (there are now 4x as many available nodes), and the ability to run larger IB jobs.

    The temporary mpirun settings that were recommended before for multinode ethernet runs, are no longer in effect, as all MPI traffic is now going over InfiniBand. For most cases, “mpirun -np X ” will work. For more details on running mpi jobs specifically on, for instance, qdr-infiniband nodes, see the wiki page on ‘GPC MPI Versions.’

    We are very interested in learning about your experiences (positive or negative) with the new infiniband network, which you can email to support@scinet.utoronto.ca.

EVENTS COMING UP

  • May 8-9: SCHEDULED SHUTDOWN OF ALL SCINET SYSTEMS

    There will be a full SciNet shutdown from Tue May 8 to Wed May 9 for final configurations in the changeover to full infiniband for the GPC, for some back-end maintenance, and to test the computational and file system performance of the TCS and GPC.

    Systems will go down at 9:00 am on May 8; all login sessions and jobs will be killed at that time. The system should be available again in the evening of the next day. Check the wiki on Wednesday for updates.

  • Wed May 9, 10:30 am – 12:00 noon: INTRO TO SCINET

    Learn what SciNet resources are available, how to compile your code and how to use the batch system, in approximately 90 minutes.

    Intended for new users, but experienced users may still pick up some valuable pointers.

    Sign up at https://support.scinet.utoronto.ca/courses/?q=node/65

    Note that attendants to the Intro may be interested in the immediately following event:

  • Wed May 9, 12:00 noon – 1:00 pm: MAY SCINET USER GROUP (SNUG) MEETING

    The SciNet Users Group (SNUG) meetings are every month on the second Wednesday, and involve pizza, user discussion, feedback, and one or two short talks on topics or technologies of interest to the SciNet community.

    This time, we will have:

    • TechTalk by Jonathan Dursi on “Parallel I/O doesn’t have to be so hard: The ADIOS library”
    • User discussion
    • Pizza!

    Sign up at: https://support.scinet.utoronto.ca/courses/?q=node/51

  • May 14-18: SCICOMP

    SciNet will host the annual meeting of ScicomP, the IBM HPC Systems Scientific Computing User Group. This meeting (which is part of the meeting of SPXXL, the user group of large IBM installations) is open to users and deals mostly with applications and science rather than just with technical aspects of the computers.

    This event will not be held at the SciNet Headquarters. For more information on this event, its schedule, location and registration, go to http://spscicomp.org/scicomp2012 .

  • Wed Jun 13, 12:00 noon – 1:00 pm: JUNE SNUG MEETING

    • TechTalk by Ramses van Zon on “Remote development on SciNet systems”
    • User discussion
    • Pizza!

    Sign up at https://support.scinet.utoronto.ca/courses/?q=node/66

ADDED TO THE WIKI IN APRIL

All new wiki content below is listed and linked on the main page: http://wiki.scinethpc.ca/wiki/index.php/SciNet_User_Support_Library#What.27s_New_On_The_Wiki)

  • Updates to all things pertaining to MPI on the GPC, in light of its new Infiniband network, including

    • Updated User Tutorial
    • Updated Scheduler Information
    • Updated GPC Cluster Information
  • Slides of the one-day c++ course.
  • A FAQ entry on how to deal with ib memory problems
  • Slides of the TechTalk on “Infiniband on the GPC”

WHAT ELSE HAPPENED AT SCINET IN APRIL?

  • Apr 11: SNUG meeting was held, with a TechTalk by Scott Northrup on “Infiniband on the GPC”
  • Apr 12: Scheduled shutdown on the TCS
  • Apr 18-19: Scheduled shutdown of all SciNet Systems for preparations for the new Blue Gene/Q system
  • Apr 19: GPC upgraded to full Infiniband
  • Apr 23: One day course on Scientific C++