PLDI 2016 (series) / X10 (series) / X10 2016 /
A Case for Distributed Work-Stealing in Regular Applications
Tue 14 Jun 2016 16:20 - 16:40 at Santa Rosa East - Research Papers
This paper presents a dynamically heterogeneous architecture use-case that is both realistic and favorable for distributed work-stealing in regular parallel applications. Using a straightforward implementation of distributed dense matrix multiplication in X10’s Global Load Balancing (GLB) library, we show that moderate differences in node processing power allow work-stealing to significantly outperform a standard static schedule such as SUMMA. It also scales comparably on up to 128 cores.
Tue 14 Jun Times are displayed in time zone: Tijuana, Baja California change
Tue 14 Jun
Times are displayed in time zone: Tijuana, Baja California change
15:30 - 16:00 Talk | ActorX10: An Actor Library for X10 X10 A: Sascha Roloff, A: Alexander Pöppl, A: Tobias Schwarzer, A: Stefan Wildermann, A: Michael Bader, A: Michael Glaß, A: Frank Hannig, A: Jürgen Teich Link to publication DOI File Attached | ||
16:00 - 16:20 Talk | SWE-X10: An Actor-Based and Locally Coordinated Solver for the Shallow Water Equations X10 Link to publication DOI File Attached | ||
16:20 - 16:40 Talk | A Case for Distributed Work-Stealing in Regular Applications X10 Link to publication DOI File Attached | ||
16:40 - 17:00 Day closing | Closing Remarks X10 |