Interval arithmetic and computational science: Performance considerations

Alistair P. Rondell; Bill Clarke; Josh Milthorpe

Interval arithmetic and computational science: Performance considerations

Alistair P. Rondell^*, Bill Clarke, Josh Milthorpe

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › peer-review

1 Citation (Scopus)

Abstract

Interval analysis is an alternative to conventional floating-point computations that offers guaranteed error bounds. Despite this advantage, interval methods have not gained widespread use in large scale computational science applications. This paper addresses this issue from a performance perspective, comparing the performance of floating point and interval operations for some small computational kernels. Particularly attention is given to the Sun Fortran interval implementation, although the strategies introduced here to enhance performance are applicable to other interval implementations. Fundamental differences in the operation counts and memory references requirements of interval and floating point codes are discussed.

Original language	English
Title of host publication	Computational Science - ICCS 2006
Subtitle of host publication	6th International Conference, Proceedings
Publisher	Springer Verlag
Pages	218-225
Number of pages	8
ISBN (Print)	3540343792, 9783540343790
Publication status	Published - 2006
Event	ICCS 2006: 6th International Conference on Computational Science - Reading, United Kingdom Duration: 28 May 2006 → 31 May 2006

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	3991 LNCS - I
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	28/05/06 → 31/05/06

Cite this

Rondell, A. P., Clarke, B., & Milthorpe, J. (2006). Interval arithmetic and computational science: Performance considerations. In Computational Science - ICCS 2006: 6th International Conference, Proceedings (pp. 218-225). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3991 LNCS - I). Springer Verlag.

Rondell, Alistair P. ; Clarke, Bill ; Milthorpe, Josh. / Interval arithmetic and computational science : Performance considerations. Computational Science - ICCS 2006: 6th International Conference, Proceedings. Springer Verlag, 2006. pp. 218-225 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c06b414776bc4d0ebe6adad4ea4c5f12,

title = "Interval arithmetic and computational science: Performance considerations",

abstract = "Interval analysis is an alternative to conventional floating-point computations that offers guaranteed error bounds. Despite this advantage, interval methods have not gained widespread use in large scale computational science applications. This paper addresses this issue from a performance perspective, comparing the performance of floating point and interval operations for some small computational kernels. Particularly attention is given to the Sun Fortran interval implementation, although the strategies introduced here to enhance performance are applicable to other interval implementations. Fundamental differences in the operation counts and memory references requirements of interval and floating point codes are discussed.",

author = "Rondell, \{Alistair P.\} and Bill Clarke and Josh Milthorpe",

year = "2006",

language = "English",

isbn = "3540343792",

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

publisher = "Springer Verlag",

pages = "218--225",

booktitle = "Computational Science - ICCS 2006",

address = "Germany",

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

}

Rondell, AP, Clarke, B & Milthorpe, J 2006, Interval arithmetic and computational science: Performance considerations. 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. 3991 LNCS - I, Springer Verlag, pp. 218-225, ICCS 2006: 6th International Conference on Computational Science, Reading, United Kingdom, 28/05/06.

Interval arithmetic and computational science: Performance considerations. / Rondell, Alistair P.; Clarke, Bill; Milthorpe, Josh.
Computational Science - ICCS 2006: 6th International Conference, Proceedings. Springer Verlag, 2006. p. 218-225 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3991 LNCS - I).

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › peer-review

TY - GEN

T1 - Interval arithmetic and computational science

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

AU - Rondell, Alistair P.

AU - Clarke, Bill

AU - Milthorpe, Josh

PY - 2006

Y1 - 2006

N2 - Interval analysis is an alternative to conventional floating-point computations that offers guaranteed error bounds. Despite this advantage, interval methods have not gained widespread use in large scale computational science applications. This paper addresses this issue from a performance perspective, comparing the performance of floating point and interval operations for some small computational kernels. Particularly attention is given to the Sun Fortran interval implementation, although the strategies introduced here to enhance performance are applicable to other interval implementations. Fundamental differences in the operation counts and memory references requirements of interval and floating point codes are discussed.

AB - Interval analysis is an alternative to conventional floating-point computations that offers guaranteed error bounds. Despite this advantage, interval methods have not gained widespread use in large scale computational science applications. This paper addresses this issue from a performance perspective, comparing the performance of floating point and interval operations for some small computational kernels. Particularly attention is given to the Sun Fortran interval implementation, although the strategies introduced here to enhance performance are applicable to other interval implementations. Fundamental differences in the operation counts and memory references requirements of interval and floating point codes are discussed.

UR - https://www.scopus.com/pages/publications/33746639573

M3 - Conference Paper

SN - 3540343792

SN - 9783540343790

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

SP - 218

EP - 225

BT - Computational Science - ICCS 2006

PB - Springer Verlag

Y2 - 28 May 2006 through 31 May 2006

ER -

Interval arithmetic and computational science: Performance considerations

Abstract

Publication series

Conference

Other files and links

Fingerprint

Cite this