Dynamic Graph Traversals for Concurrent Rewriting Using Work-Stealing Frameworks for Multicore Platforms

Parallel programming/execution frameworks for multicore platforms should support as many applications as possible. In general, work-stealing frameworks provide efficient load balancing even for irregular parallel applications. Unfortunately, naive parallel programs for ``general computations'&#...

Full description

Saved in:
Bibliographic Details
Published in:2010 IEEE 16th International Conference on Parallel and Distributed Systems pp. 406 - 414
Main Authors: Yasugi, M, Hiraishi, T, Umatani, S, Yuasa, T
Format: Conference Proceeding
Language:English
Japanese
Published: IEEE 01.12.2010
Subjects:
graph traversals
Instruction sets
Load management
Multicore processing
multicores
Parallel programming
Proposals
Protocols
spanning tree
Synchronization
work stealing
ISBN:1424497272, 9781424497270
ISSN:1521-9097
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Parallel programming/execution frameworks for multicore platforms should support as many applications as possible. In general, work-stealing frameworks provide efficient load balancing even for irregular parallel applications. Unfortunately, naive parallel programs for ``general computations'' governed by rewriting rules over configurations (system states) cause stack overflow or unacceptable load imbalance. In this study, we develop parallel programs to perform probabilistically balanced divide-and-conquer graph traversals. We propose a programming technique for accumulating overflowed calls for the next iteration of repeated parallel stages. In an emerging backtracking-based work-stealing framework called "Tascell," which avoids excessive concurrency, we propose a programming technique for long-term use of workspaces, leading a similar technique in the Cilk framework.
ISBN:1424497272
9781424497270
ISSN:1521-9097
DOI:10.1109/ICPADS.2010.119