Floating-point to fixed-point transformation using extreme value theory

Linsheng Zhang; Yan Zhang; Wenbiao Zhou

doi:10.1109/ICIS.2009.38

Floating-point to fixed-point transformation using extreme value theory

Linsheng Zhang^*
, Yan Zhang
, Wenbiao Zhou

^*Corresponding author for this work

Harbin Institute of Technology Shenzhen

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

12 Scopus citations

Abstract

Efficient hardware implementation of digital signal processing and communication algorithms requires using fixed-point arithmetic. However, most of these algorithms are developed in floating-point arithmetic. The automation of floating-point to fixed-point transformation is a key step for high-level synthesis. Based on Extreme Value Theory, this paper presents a novel approach to find both the optimal integer and fractional bit-widths for signals. Compared to traditional full simulation-based methods, this approach uses lightweight simulations to study the characteristics of extreme conditions. With theoretical probabilities to prevent overflows and meet the output error constraint, the proposed approach provides more practical solutions than those by analytical methods. Case studies and experimental results demonstrate its efficiency.

Original language	English
Title of host publication	Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009
Publisher	IEEE Computer Society
Pages	271-276
Number of pages	6
ISBN (Print)	9780769536415
DOIs	https://doi.org/10.1109/ICIS.2009.38
State	Published - 2009
Externally published	Yes
Event	8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009 - Shanghai, China Duration: 1 Jun 2009 → 3 Jun 2009

Publication series

Name	Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009

Conference

Conference	8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009
Country/Territory	China
City	Shanghai
Period	1/06/09 → 3/06/09

Access to Document

10.1109/ICIS.2009.38

Cite this

Zhang, L., Zhang, Y., & Zhou, W. (2009). Floating-point to fixed-point transformation using extreme value theory. In Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009 (pp. 271-276). Article 5222868 (Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009). IEEE Computer Society. https://doi.org/10.1109/ICIS.2009.38

Zhang, Linsheng ; Zhang, Yan ; Zhou, Wenbiao. / Floating-point to fixed-point transformation using extreme value theory. Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009. IEEE Computer Society, 2009. pp. 271-276 (Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009).

@inproceedings{3eb884b5c44d4e3fa95efa49f60766fc,

title = "Floating-point to fixed-point transformation using extreme value theory",

abstract = "Efficient hardware implementation of digital signal processing and communication algorithms requires using fixed-point arithmetic. However, most of these algorithms are developed in floating-point arithmetic. The automation of floating-point to fixed-point transformation is a key step for high-level synthesis. Based on Extreme Value Theory, this paper presents a novel approach to find both the optimal integer and fractional bit-widths for signals. Compared to traditional full simulation-based methods, this approach uses lightweight simulations to study the characteristics of extreme conditions. With theoretical probabilities to prevent overflows and meet the output error constraint, the proposed approach provides more practical solutions than those by analytical methods. Case studies and experimental results demonstrate its efficiency.",

author = "Linsheng Zhang and Yan Zhang and Wenbiao Zhou",

year = "2009",

doi = "10.1109/ICIS.2009.38",

language = "英语",

isbn = "9780769536415",

series = "Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009",

publisher = "IEEE Computer Society",

pages = "271--276",

booktitle = "Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009",

address = "美国",

note = "8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009 ; Conference date: 01-06-2009 Through 03-06-2009",

}

Zhang, L, Zhang, Y & Zhou, W 2009, Floating-point to fixed-point transformation using extreme value theory. in Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009., 5222868, Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009, IEEE Computer Society, pp. 271-276, 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009, Shanghai, China, 1/06/09. https://doi.org/10.1109/ICIS.2009.38

Floating-point to fixed-point transformation using extreme value theory. / Zhang, Linsheng; Zhang, Yan; Zhou, Wenbiao.
Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009. IEEE Computer Society, 2009. p. 271-276 5222868 (Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009).

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

TY - GEN

T1 - Floating-point to fixed-point transformation using extreme value theory

AU - Zhang, Linsheng

AU - Zhang, Yan

AU - Zhou, Wenbiao

PY - 2009

Y1 - 2009

N2 - Efficient hardware implementation of digital signal processing and communication algorithms requires using fixed-point arithmetic. However, most of these algorithms are developed in floating-point arithmetic. The automation of floating-point to fixed-point transformation is a key step for high-level synthesis. Based on Extreme Value Theory, this paper presents a novel approach to find both the optimal integer and fractional bit-widths for signals. Compared to traditional full simulation-based methods, this approach uses lightweight simulations to study the characteristics of extreme conditions. With theoretical probabilities to prevent overflows and meet the output error constraint, the proposed approach provides more practical solutions than those by analytical methods. Case studies and experimental results demonstrate its efficiency.

AB - Efficient hardware implementation of digital signal processing and communication algorithms requires using fixed-point arithmetic. However, most of these algorithms are developed in floating-point arithmetic. The automation of floating-point to fixed-point transformation is a key step for high-level synthesis. Based on Extreme Value Theory, this paper presents a novel approach to find both the optimal integer and fractional bit-widths for signals. Compared to traditional full simulation-based methods, this approach uses lightweight simulations to study the characteristics of extreme conditions. With theoretical probabilities to prevent overflows and meet the output error constraint, the proposed approach provides more practical solutions than those by analytical methods. Case studies and experimental results demonstrate its efficiency.

UR - https://www.scopus.com/pages/publications/70350736312

U2 - 10.1109/ICIS.2009.38

DO - 10.1109/ICIS.2009.38

M3 - 会议稿件

AN - SCOPUS:70350736312

SN - 9780769536415

T3 - Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009

SP - 271

EP - 276

BT - Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009

PB - IEEE Computer Society

T2 - 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009

Y2 - 1 June 2009 through 3 June 2009

ER -

Zhang L, Zhang Y, Zhou W. Floating-point to fixed-point transformation using extreme value theory. In Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009. IEEE Computer Society. 2009. p. 271-276. 5222868. (Proceedings of the 2009 8th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2009). doi: 10.1109/ICIS.2009.38

Floating-point to fixed-point transformation using extreme value theory

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this