Loading...
Loading...

Go to the content (press return)

Introducing speculative optimizations in task dataflow with language extensions and runtime support

Author
Azuelos, N.; Etsion, Y.; Keidar, I.; Zaks, A.; Ayguade, E.
Type of activity
Presentation of work at congresses
Name of edition
2nd Workshop on Data-Flow Execution Models for Extreme Scale Computing (DFM 2012)
Date of publication
2012
Presentation's date
2012-09
Book of congress proceedings
2nd Workshop on Data-Flow Execution Models for Extreme Scale Computing
First page
1
Last page
4
Repository
http://hdl.handle.net/2117/18339 Open in new window
URL
http://www.cs.ucy.ac.cy/dfmworkshop/assets/dfm2012/Introducing%20Speculative%20Optimizations%20in%20Task%20Dataflow%20with%20Language%20Extensions%20and%20Runtime%20Support.pdf Open in new window
Abstract
We argue that speculation leads to increased parallelism in the coarse-grain dataflow paradigm. To do so, we present a framework for adding speculation in a popular and well-established framework. We specify a limited set of additions to the OmpSs language and changes required in its supporting runtime environment. These modifications enable speculation across the system in a flexible way. We evaluate our implementation using a simple benchmark leading to a promising 10% speedup.
Citation
Azuelos, N. [et al.]. Introducing speculative optimizations in task dataflow with language extensions and runtime support. A: Workshop on Data-Flow Execution Models for Extreme Scale Computing. "2nd Workshop on Data-Flow Execution Models for Extreme Scale Computing". Minneapolis: 2012, p. 1-4.
Group of research
CAP - High Performace Computing Group

Participants

  • Azuelos, Nathaniel  (author and speaker )
  • Etsion, Yoav  (author and speaker )
  • Keidar, Idit  (author and speaker )
  • Zaks, A.  (author and speaker )
  • Ayguade Parra, Eduard  (author and speaker )