File Download
  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: Adaptive thread scheduling techniques for improving scalability of software transactional memory

TitleAdaptive thread scheduling techniques for improving scalability of software transactional memory
Authors
KeywordsAdaptive concurrency control
Software transactional memory
Thread scheduling
Issue Date2011
PublisherACTA Press.
Citation
The 10th IASTED International Conference on Parallel and Distributed Computing and Networks (PDCN 2011), Innsbruck, Austria, 15-17 February 2011. In Proceedings of the 10th IASTED-PDCN, 2011, p. 91-98 How to Cite?
AbstractSoftware transactional memory (STM) enhances both ease-of-use and concurrency, and is considered state-of-the-art for parallel applications to scale on modern multi-core hardware. However, there are certain situations where STM performs even worse than traditional locks. Upon hotspots where most threads contend over a few pieces of shared data, going transactional will result in excessive conflicts and aborts that adversely degrade performance. We present a new design of adaptive thread scheduler that manages concurrency when the system is about entering and leaving hotspots. The scheduler controls the number of threads spawning new transactions according to the live commit throughput. We implemented two feedback-control policies called Throttle and Probe to realize this adaptive scheduling. Performance evaluation with the STAMP benchmarks shows that enabling Throttle and Probe obtain best-case speedups of 87.5% and 108.7% respectively.
Persistent Identifierhttp://hdl.handle.net/10722/144628
ISBN
References

 

DC FieldValueLanguage
dc.contributor.authorChan, Ken_HK
dc.contributor.authorLam, KTen_HK
dc.contributor.authorWang, CLen_HK
dc.date.accessioned2012-02-03T06:16:59Z-
dc.date.available2012-02-03T06:16:59Z-
dc.date.issued2011en_HK
dc.identifier.citationThe 10th IASTED International Conference on Parallel and Distributed Computing and Networks (PDCN 2011), Innsbruck, Austria, 15-17 February 2011. In Proceedings of the 10th IASTED-PDCN, 2011, p. 91-98en_HK
dc.identifier.isbn9780889868649-
dc.identifier.urihttp://hdl.handle.net/10722/144628-
dc.description.abstractSoftware transactional memory (STM) enhances both ease-of-use and concurrency, and is considered state-of-the-art for parallel applications to scale on modern multi-core hardware. However, there are certain situations where STM performs even worse than traditional locks. Upon hotspots where most threads contend over a few pieces of shared data, going transactional will result in excessive conflicts and aborts that adversely degrade performance. We present a new design of adaptive thread scheduler that manages concurrency when the system is about entering and leaving hotspots. The scheduler controls the number of threads spawning new transactions according to the live commit throughput. We implemented two feedback-control policies called Throttle and Probe to realize this adaptive scheduling. Performance evaluation with the STAMP benchmarks shows that enabling Throttle and Probe obtain best-case speedups of 87.5% and 108.7% respectively.en_HK
dc.languageengen_US
dc.publisherACTA Press.-
dc.relation.ispartofProceedings of the 10th IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2011en_HK
dc.rightsCreative Commons: Attribution 3.0 Hong Kong License-
dc.subjectAdaptive concurrency controlen_HK
dc.subjectSoftware transactional memoryen_HK
dc.subjectThread schedulingen_HK
dc.titleAdaptive thread scheduling techniques for improving scalability of software transactional memoryen_HK
dc.typeConference_Paperen_HK
dc.identifier.emailChan, K: kchan@cs.hku.hken_HK
dc.identifier.emailLam, KT: ktlam@cs.hku.hk-
dc.identifier.emailWang, CL: clwang@cs.hku.hk-
dc.identifier.authorityWang, CL=rp00183en_HK
dc.description.naturepostprint-
dc.identifier.doi10.2316/P.2011.719-011en_HK
dc.identifier.scopuseid_2-s2.0-79958078740en_HK
dc.identifier.hkuros198271en_US
dc.relation.referenceshttp://www.scopus.com/mlt/select.url?eid=2-s2.0-79958078740&selection=ref&src=s&origin=recordpageen_HK
dc.identifier.spage91en_HK
dc.identifier.epage98en_HK
dc.description.otherThe 10th IASTED International Conference on Parallel and Distributed Computing and Networks (PDCN 2011), Innsbruck, Austria, 15-17 February 2011. In Proceedings of the 10th IASTED-PDCN, 2011, p. 91-98-
dc.identifier.scopusauthoridWang, CL=7501646188en_HK
dc.identifier.scopusauthoridLam, KT=26031004100en_HK
dc.identifier.scopusauthoridChan, K=36915734900en_HK

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats