Exploring the possibility of operating in the compressed domain

V. De La Luz; M. Kandemir; A. Sivasubramaniam; M. J. Irwin

doi:10.1007/978-3-540-27866-5_67

Exploring the possibility of operating in the compressed domain

V. De La Luz
, M. Kandemir
, A. Sivasubramaniam
, M. J. Irwin

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

Abstract

Data compression in caches has been studied from the performance and energy consumption points of view. In this paper, we study the possible benefits of operating with compressed operands. Operating in the compressed domain (i.e., executing instructions with compressed operands) as far as possible can offer several advantages. First, since it is less reliant on decompression, it may be possible to employ a fancier compression/decompression strategy, without the associated performance/power penalties, that can offer higher compression rates to further boost cache locality. Second, the transfers between cache and datapath can use fewer bits (to transmit codes instead of data values) and provide dynamic energy savings in the corresponding bus. In this paper, we demonstrate that there are cases where operations can be executed in the compressed domain, which can lead to performance improvements and energy savings. In particular, we show that by operating in the compressed domain, the effectiveness of prior techniques can be further improved.

Original language	English (US)
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Editors	Marco Danelutto, Marco Vanneschi, Domenico Laforenza
Publisher	Springer Verlag
Pages	507-515
Number of pages	9
ISBN (Print)	3540229248
DOIs	https://doi.org/10.1007/978-3-540-27866-5_67
State	Published - 2004

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	3149
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-540-27866-5_67

Cite this

De La Luz, V., Kandemir, M., Sivasubramaniam, A., & Irwin, M. J. (2004). Exploring the possibility of operating in the compressed domain. In M. Danelutto, M. Vanneschi, & D. Laforenza (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 507-515). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3149). Springer Verlag. https://doi.org/10.1007/978-3-540-27866-5_67

De La Luz, V. ; Kandemir, M. ; Sivasubramaniam, A. et al. / Exploring the possibility of operating in the compressed domain. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). editor / Marco Danelutto ; Marco Vanneschi ; Domenico Laforenza. Springer Verlag, 2004. pp. 507-515 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inbook{60d79b6d4cc047f18c6ebaf1a3f51cad,

title = "Exploring the possibility of operating in the compressed domain",

abstract = "Data compression in caches has been studied from the performance and energy consumption points of view. In this paper, we study the possible benefits of operating with compressed operands. Operating in the compressed domain (i.e., executing instructions with compressed operands) as far as possible can offer several advantages. First, since it is less reliant on decompression, it may be possible to employ a fancier compression/decompression strategy, without the associated performance/power penalties, that can offer higher compression rates to further boost cache locality. Second, the transfers between cache and datapath can use fewer bits (to transmit codes instead of data values) and provide dynamic energy savings in the corresponding bus. In this paper, we demonstrate that there are cases where operations can be executed in the compressed domain, which can lead to performance improvements and energy savings. In particular, we show that by operating in the compressed domain, the effectiveness of prior techniques can be further improved.",

author = "\{De La Luz\}, V. and M. Kandemir and A. Sivasubramaniam and Irwin, \{M. J.\}",

year = "2004",

doi = "10.1007/978-3-540-27866-5\_67",

language = "English (US)",

isbn = "3540229248",

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

publisher = "Springer Verlag",

pages = "507--515",

editor = "Marco Danelutto and Marco Vanneschi and Domenico Laforenza",

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

address = "Germany",

}

De La Luz, V, Kandemir, M , Sivasubramaniam, A & Irwin, MJ 2004, Exploring the possibility of operating in the compressed domain. in M Danelutto, M Vanneschi & D Laforenza (eds), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3149, Springer Verlag, pp. 507-515. https://doi.org/10.1007/978-3-540-27866-5_67

Exploring the possibility of operating in the compressed domain. / De La Luz, V.; Kandemir, M.; Sivasubramaniam, A. et al.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). ed. / Marco Danelutto; Marco Vanneschi; Domenico Laforenza. Springer Verlag, 2004. p. 507-515 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3149).

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

TY - CHAP

T1 - Exploring the possibility of operating in the compressed domain

AU - De La Luz, V.

AU - Kandemir, M.

AU - Sivasubramaniam, A.

AU - Irwin, M. J.

PY - 2004

Y1 - 2004

N2 - Data compression in caches has been studied from the performance and energy consumption points of view. In this paper, we study the possible benefits of operating with compressed operands. Operating in the compressed domain (i.e., executing instructions with compressed operands) as far as possible can offer several advantages. First, since it is less reliant on decompression, it may be possible to employ a fancier compression/decompression strategy, without the associated performance/power penalties, that can offer higher compression rates to further boost cache locality. Second, the transfers between cache and datapath can use fewer bits (to transmit codes instead of data values) and provide dynamic energy savings in the corresponding bus. In this paper, we demonstrate that there are cases where operations can be executed in the compressed domain, which can lead to performance improvements and energy savings. In particular, we show that by operating in the compressed domain, the effectiveness of prior techniques can be further improved.

AB - Data compression in caches has been studied from the performance and energy consumption points of view. In this paper, we study the possible benefits of operating with compressed operands. Operating in the compressed domain (i.e., executing instructions with compressed operands) as far as possible can offer several advantages. First, since it is less reliant on decompression, it may be possible to employ a fancier compression/decompression strategy, without the associated performance/power penalties, that can offer higher compression rates to further boost cache locality. Second, the transfers between cache and datapath can use fewer bits (to transmit codes instead of data values) and provide dynamic energy savings in the corresponding bus. In this paper, we demonstrate that there are cases where operations can be executed in the compressed domain, which can lead to performance improvements and energy savings. In particular, we show that by operating in the compressed domain, the effectiveness of prior techniques can be further improved.

UR - https://www.scopus.com/pages/publications/35048824547

UR - https://www.scopus.com/pages/publications/35048824547#tab=citedBy

U2 - 10.1007/978-3-540-27866-5_67

DO - 10.1007/978-3-540-27866-5_67

M3 - Chapter

AN - SCOPUS:35048824547

SN - 3540229248

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

SP - 507

EP - 515

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

A2 - Danelutto, Marco

A2 - Vanneschi, Marco

A2 - Laforenza, Domenico

PB - Springer Verlag

ER -

De La Luz V, Kandemir M , Sivasubramaniam A, Irwin MJ. Exploring the possibility of operating in the compressed domain. In Danelutto M, Vanneschi M, Laforenza D, editors, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Springer Verlag. 2004. p. 507-515. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-540-27866-5_67

Exploring the possibility of operating in the compressed domain

Abstract

Publication series

UN SDGs

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this