Introductory: 10% Intermediate: 50% Advanced: 40%

SC Conference - Activity Details

M05: Hybrid Parallel Programming and Multi-Core Optimization on the Ranger Supercomputer

Dan C. Stanzione  (Arizona State University)
Byoung-Do Kim  (Texas Advanced Computing Center)
Kent Milfield  (Texas Advanced Computing Center)
Gabriele Jost  (Texas Advanced Computing Center)
Tutorials Session
Monday,  08:30AM - 05:00PM
Room TBD
The Ranger supercomputer is the first system deployed through the NSF Track 2, "Path to Petascale" program. This system offers more than a half petaflop of peak performance and unprecedented capability and potential to run scalable applications. Ranger is based on quad-core AMD processors in quad-socket blades from Sun Microsystems, with a total of 62,976 cores in the system. These 16 core nodes, combined with the total scale of the system, provide challenges in effectively designing and developing large scale applications. This tutorial targets users with experience in parallel computing (specifically, familiarity with MPI) who wish to migrate to like Ranger. In the morning, the Ranger user environment will be presented along with performance information, experiences and evaluations for programming in a hybrid mode. In the afternoon, the tutorial delves into performance optimization on multi-core nodes, including NUMA tools. Best practices �templates� introducing hybrid computing into applications will be provided.
   IEEE Computer Society  /  ACM     2 0   Y E A R S   -   U N L E A S H I N G   T H E   P O W E R   O F   H P C