Abstract
This paper introduces a new mechanism for the exposure of large grain parallelism. The scheme performs lazy task creation; inlining all tasks provisionally and extracting parallelism from the inlined information later on demand. However, unlike other mechanisms, the further task demand is satisfied by the next evaluation stream rather than retrospectively reversing the inlining decision of the current stream. The scheme is called lazy decomposition because decomposition itself is throttled rather than just the extraction of a task. Lazy decomposition makes the serial section clearly separated from the parallel section in an evaluation tree for a particular function, and this allows the serial section to adopt a sequential algorithm. The performance improvement is significant in divide-and-conquer applications by adoption of sequential algorithms.
Original language | English |
---|---|
Title of host publication | 1997 3rd International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 1997 |
Editors | Wanlei Zhou, Andrzej Goscinski, Michael Hobbs |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
Pages | 427-434 |
Number of pages | 8 |
ISBN (Electronic) | 0780342291, 9780780342293 |
DOIs | |
Publication status | Published - 1997 |
Event | 3rd International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 1997 - Melbourne, Australia Duration: 1997 Dec 10 → 1997 Dec 12 |
Publication series
Name | 1997 3rd International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 1997 |
---|
Other
Other | 3rd International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 1997 |
---|---|
Country/Territory | Australia |
City | Melbourne |
Period | 97/12/10 → 97/12/12 |
Bibliographical note
Publisher Copyright:© 1997 IEEE.
All Science Journal Classification (ASJC) codes
- Computer Networks and Communications
- Hardware and Architecture
- Signal Processing