SC is the International Conference for
 High Performnance Computing, Networking, Storage and Analysis

SCHEDULE: NOV 13-19, 2010

Comparing Parallel Programming Models for Random Remote Updates

SESSION: Student Poster Reception

EVENT TYPE: Poster, ACM Student Poster

TIME: 5:15PM - 7:00PM

AUTHOR(S):Van Bui

ROOM:Main Lobby

ABSTRACT:
It can be argued that one source of productivity problems in parallel programming is the need to aggregate remote data operations in order to achieve (or preserve) performance. In this work, we compare simple variations of the random access benchmark from the HPC Challenge Benchmarks with a variety of parallel programming models to evaluate both the state of the implementations of those models and to discuss how much of the difference between the models is due to differences in the abilities of the programming models and how much reflects the state of the implementations. We develop a performance model to evaluate the costs of local and remote updates across different programming models and different implementations of the programming models. Based on our models and validation, we observe that the synchronization overhead in all our implemented versions is the main factor limiting scalability of RandomAccess.

Chair/Author Details:

Van Bui - University of Illinois at Urbana-Champaign

Add to iCal  Click here to download .ics calendar file

Add to Outlook  Click here to download .vcs calendar file

Add to Google Calendarss  Click here to add event to your Google Calendar

   Sponsors    IEEE    ACM