Hybrid static/dynamic activity analysis

Barbara Kreaseck; Luis Ramos; Scott Easterday; Michelle Strout; Paul Hovland

doi:10.1007/11758549_80

Hybrid static/dynamic activity analysis

Barbara Kreaseck, Luis Ramos, Scott Easterday, Michelle Strout, Paul Hovland

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

3 Scopus citations

Abstract

In forward mode Automatic Differentiation, the derivative program computes a function f and its derivatives, f′. Activity analysis is important for AD. Our results show that when all variables are active, the runtime checks required for dynamic activity analysis incur a significant overhead. However, when as few as half of the input variables are inactive, dynamic activity analysis enables an average speedup of 28% on a set of benchmark problems. We investigate static activity analysis combined with dynamic activity analysis as a technique for reducing the overhead of dynamic activity analysis.

Original language	English (US)
Title of host publication	Computational Science - ICCS 2006
Subtitle of host publication	6th International Conference, Proceedings
Publisher	Springer-Verlag
Pages	582-590
Number of pages	9
ISBN (Print)	3540343857, 9783540343851
DOIs	https://doi.org/10.1007/11758549_80
State	Published - 2006
Externally published	Yes
Event	ICCS 2006: 6th International Conference on Computational Science - Reading, United Kingdom Duration: May 28 2006 → May 31 2006

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	3994 LNCS - IV
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	ICCS 2006: 6th International Conference on Computational Science
Country/Territory	United Kingdom
City	Reading
Period	5/28/06 → 5/31/06

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/11758549_80

Cite this

Kreaseck, B., Ramos, L., Easterday, S., Strout, M., & Hovland, P. (2006). Hybrid static/dynamic activity analysis. In Computational Science - ICCS 2006: 6th International Conference, Proceedings (pp. 582-590). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3994 LNCS - IV). Springer-Verlag. https://doi.org/10.1007/11758549_80

Hybrid static/dynamic activity analysis. / Kreaseck, Barbara; Ramos, Luis; Easterday, Scott et al.

Computational Science - ICCS 2006: 6th International Conference, Proceedings. Springer-Verlag, 2006. p. 582-590 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3994 LNCS - IV).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Kreaseck, B, Ramos, L, Easterday, S, Strout, M & Hovland, P 2006, Hybrid static/dynamic activity analysis. in Computational Science - ICCS 2006: 6th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3994 LNCS - IV, Springer-Verlag, pp. 582-590, ICCS 2006: 6th International Conference on Computational Science, Reading, United Kingdom, 5/28/06. https://doi.org/10.1007/11758549_80

Kreaseck B, Ramos L, Easterday S, Strout M, Hovland P. Hybrid static/dynamic activity analysis. In Computational Science - ICCS 2006: 6th International Conference, Proceedings. Springer-Verlag. 2006. p. 582-590. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/11758549_80

@inproceedings{5bd59576a7bb4c598248636bd39c8528,

title = "Hybrid static/dynamic activity analysis",

abstract = "In forward mode Automatic Differentiation, the derivative program computes a function f and its derivatives, f′. Activity analysis is important for AD. Our results show that when all variables are active, the runtime checks required for dynamic activity analysis incur a significant overhead. However, when as few as half of the input variables are inactive, dynamic activity analysis enables an average speedup of 28% on a set of benchmark problems. We investigate static activity analysis combined with dynamic activity analysis as a technique for reducing the overhead of dynamic activity analysis.",

author = "Barbara Kreaseck and Luis Ramos and Scott Easterday and Michelle Strout and Paul Hovland",

year = "2006",

doi = "10.1007/11758549_80",

language = "English (US)",

isbn = "3540343857",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer-Verlag",

pages = "582--590",

booktitle = "Computational Science - ICCS 2006",

note = "ICCS 2006: 6th International Conference on Computational Science ; Conference date: 28-05-2006 Through 31-05-2006",

}

TY - GEN

T1 - Hybrid static/dynamic activity analysis

AU - Kreaseck, Barbara

AU - Ramos, Luis

AU - Easterday, Scott

AU - Strout, Michelle

AU - Hovland, Paul

PY - 2006

Y1 - 2006

N2 - In forward mode Automatic Differentiation, the derivative program computes a function f and its derivatives, f′. Activity analysis is important for AD. Our results show that when all variables are active, the runtime checks required for dynamic activity analysis incur a significant overhead. However, when as few as half of the input variables are inactive, dynamic activity analysis enables an average speedup of 28% on a set of benchmark problems. We investigate static activity analysis combined with dynamic activity analysis as a technique for reducing the overhead of dynamic activity analysis.

AB - In forward mode Automatic Differentiation, the derivative program computes a function f and its derivatives, f′. Activity analysis is important for AD. Our results show that when all variables are active, the runtime checks required for dynamic activity analysis incur a significant overhead. However, when as few as half of the input variables are inactive, dynamic activity analysis enables an average speedup of 28% on a set of benchmark problems. We investigate static activity analysis combined with dynamic activity analysis as a technique for reducing the overhead of dynamic activity analysis.

UR - http://www.scopus.com/inward/record.url?scp=33746630471&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33746630471&partnerID=8YFLogxK

U2 - 10.1007/11758549_80

DO - 10.1007/11758549_80

M3 - Conference contribution

AN - SCOPUS:33746630471

SN - 3540343857

SN - 9783540343851

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 582

EP - 590

BT - Computational Science - ICCS 2006

PB - Springer-Verlag

T2 - ICCS 2006: 6th International Conference on Computational Science

Y2 - 28 May 2006 through 31 May 2006

ER -

Hybrid static/dynamic activity analysis

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this