SuperMUC Training: Using the GPFS parallel file system

[LRZ aktuell]

Dear users of the supzero migration system,
 
 When user operation of the Petaflop system "SuperMUC" starts in summer
 2012, a new and very powerful I/O subsystem will also become available.
 This will allow large-scale parallel I/O with an aggregate bandwidth of
 200 GB/s and be based on IBM's General Parallel File system (GPFS).
 
 Handling large-scale I/O is not a trivial task and requires some
 knowledge of the properties of GPFS and its tools and facilities for I/
 O tuning. In order to provide this knowledge, a one-day workshop will
 take place on January 30, 2012; participation in this event is highly
 recommended for all SuperMUC users who do large-scale I/O (either many
 small files, or any number of large files) with their simulation
 programs.
 
 Further details are available at http://www.lrz.de/services/compute/
 courses/#GPFSTraining.
 
 Registration for the event is required since only a limited number of
 seats is available.


 This information is also available on our web server
 http://www.lrz-muenchen.de/services/compute/supermuc/aktuell/ali4182/

 Reinhold Bader