Improving offset assignment on embedded processors using transformations

Sunil Atri, J. Ramanujam, Mahmut Kandemir

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Citations (Scopus)

Abstract

Embedded systems consisting of the application program ROM, RAM, the embedded processor core and any custom hardware on a single wafer are becoming increasingly common in areas such as signal processing. In this paper, we address new code optimization techniques for embedded fixed point DSP processors which have limited on-chip program ROM and include indirect addressing modes usingpost increment and decrement operations. These addressingmodes allow for efficient sequential access but the addressinginstructions increase code size. Most of the previous approaches to the problem aim to find a placement or layout of variables in the memory so that it is possible to subsume explicit address pointer manipulation instructions into other instructions as a post-increment or post-decrement operation. Our solution is aimed at transformingthe access pattern by usingproperties of operators such as commutativity so that current algorithms for variable placement are more effective.

Original languageEnglish (US)
Title of host publicationHigh Performance Computing - HiPC 2000 - 7th International Conference, Proceedings
EditorsMateo Valero, Sriram Vajapeyam, Viktor K. Prasanna
PublisherSpringer Verlag
Pages367-374
Number of pages8
ISBN (Print)3540414290, 9783540414292
StatePublished - Jan 1 2000
Event7th International Conference on High Performance Computing, HiPC 2000 - Bangalore, India
Duration: Dec 17 2000Dec 20 2000

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1970
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other7th International Conference on High Performance Computing, HiPC 2000
CountryIndia
CityBangalore
Period12/17/0012/20/00

Fingerprint

Embedded Processor
ROM
Increment
Placement
Assignment
Commutativity
Random access storage
Application programs
Embedded systems
Embedded Systems
Wafer
Optimization Techniques
Signal Processing
Mathematical operators
Layout
Manipulation
Signal processing
Chip
Fixed point
Hardware

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Atri, S., Ramanujam, J., & Kandemir, M. (2000). Improving offset assignment on embedded processors using transformations. In M. Valero, S. Vajapeyam, & V. K. Prasanna (Eds.), High Performance Computing - HiPC 2000 - 7th International Conference, Proceedings (pp. 367-374). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1970). Springer Verlag.
Atri, Sunil ; Ramanujam, J. ; Kandemir, Mahmut. / Improving offset assignment on embedded processors using transformations. High Performance Computing - HiPC 2000 - 7th International Conference, Proceedings. editor / Mateo Valero ; Sriram Vajapeyam ; Viktor K. Prasanna. Springer Verlag, 2000. pp. 367-374 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{ec1217e6e40945e6a4ef4e251f95e32a,
title = "Improving offset assignment on embedded processors using transformations",
abstract = "Embedded systems consisting of the application program ROM, RAM, the embedded processor core and any custom hardware on a single wafer are becoming increasingly common in areas such as signal processing. In this paper, we address new code optimization techniques for embedded fixed point DSP processors which have limited on-chip program ROM and include indirect addressing modes usingpost increment and decrement operations. These addressingmodes allow for efficient sequential access but the addressinginstructions increase code size. Most of the previous approaches to the problem aim to find a placement or layout of variables in the memory so that it is possible to subsume explicit address pointer manipulation instructions into other instructions as a post-increment or post-decrement operation. Our solution is aimed at transformingthe access pattern by usingproperties of operators such as commutativity so that current algorithms for variable placement are more effective.",
author = "Sunil Atri and J. Ramanujam and Mahmut Kandemir",
year = "2000",
month = "1",
day = "1",
language = "English (US)",
isbn = "3540414290",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "367--374",
editor = "Mateo Valero and Sriram Vajapeyam and Prasanna, {Viktor K.}",
booktitle = "High Performance Computing - HiPC 2000 - 7th International Conference, Proceedings",
address = "Germany",

}

Atri, S, Ramanujam, J & Kandemir, M 2000, Improving offset assignment on embedded processors using transformations. in M Valero, S Vajapeyam & VK Prasanna (eds), High Performance Computing - HiPC 2000 - 7th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1970, Springer Verlag, pp. 367-374, 7th International Conference on High Performance Computing, HiPC 2000, Bangalore, India, 12/17/00.

Improving offset assignment on embedded processors using transformations. / Atri, Sunil; Ramanujam, J.; Kandemir, Mahmut.

High Performance Computing - HiPC 2000 - 7th International Conference, Proceedings. ed. / Mateo Valero; Sriram Vajapeyam; Viktor K. Prasanna. Springer Verlag, 2000. p. 367-374 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1970).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Improving offset assignment on embedded processors using transformations

AU - Atri, Sunil

AU - Ramanujam, J.

AU - Kandemir, Mahmut

PY - 2000/1/1

Y1 - 2000/1/1

N2 - Embedded systems consisting of the application program ROM, RAM, the embedded processor core and any custom hardware on a single wafer are becoming increasingly common in areas such as signal processing. In this paper, we address new code optimization techniques for embedded fixed point DSP processors which have limited on-chip program ROM and include indirect addressing modes usingpost increment and decrement operations. These addressingmodes allow for efficient sequential access but the addressinginstructions increase code size. Most of the previous approaches to the problem aim to find a placement or layout of variables in the memory so that it is possible to subsume explicit address pointer manipulation instructions into other instructions as a post-increment or post-decrement operation. Our solution is aimed at transformingthe access pattern by usingproperties of operators such as commutativity so that current algorithms for variable placement are more effective.

AB - Embedded systems consisting of the application program ROM, RAM, the embedded processor core and any custom hardware on a single wafer are becoming increasingly common in areas such as signal processing. In this paper, we address new code optimization techniques for embedded fixed point DSP processors which have limited on-chip program ROM and include indirect addressing modes usingpost increment and decrement operations. These addressingmodes allow for efficient sequential access but the addressinginstructions increase code size. Most of the previous approaches to the problem aim to find a placement or layout of variables in the memory so that it is possible to subsume explicit address pointer manipulation instructions into other instructions as a post-increment or post-decrement operation. Our solution is aimed at transformingthe access pattern by usingproperties of operators such as commutativity so that current algorithms for variable placement are more effective.

UR - http://www.scopus.com/inward/record.url?scp=84947556686&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84947556686&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84947556686

SN - 3540414290

SN - 9783540414292

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

SP - 367

EP - 374

BT - High Performance Computing - HiPC 2000 - 7th International Conference, Proceedings

A2 - Valero, Mateo

A2 - Vajapeyam, Sriram

A2 - Prasanna, Viktor K.

PB - Springer Verlag

ER -

Atri S, Ramanujam J, Kandemir M. Improving offset assignment on embedded processors using transformations. In Valero M, Vajapeyam S, Prasanna VK, editors, High Performance Computing - HiPC 2000 - 7th International Conference, Proceedings. Springer Verlag. 2000. p. 367-374. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).