Performance of PGA (Programmable Graph Architecture) for matrix multiplications

Muling Peng; Sreekanth Ramani; K. Wendy Tang; A. Yavuz Oruç

doi:10.1109/ICIS-COMSAR.2006.67

Performance of PGA (Programmable Graph Architecture) for matrix multiplications

Muling Peng
, Sreekanth Ramani
, K. Wendy Tang
, A. Yavuz Oruç

Electrical Engineering

Stony Brook University
University of Maryland, College Park

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

1 Scopus citations

Abstract

Matrix multiplication is one of the primary operations in linear algebra and is useful in a wide spectrum of multimedia applications including signal and image processing. In this paper we review a novel computer architecture for matrix multiplications. This novel architecture is based on graph theory, hence the name Programmable Graph Architecture. We discuss the performance of PGA via time and space complexity and time-processor product for parallel implementation. We further compare the PGA performance with Cannon's algorithm and show that the PGA architecture outperforms the existing method for matrix with any dimension.

Original language	English
Title of host publication	Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S
Pages	361-367
Number of pages	7
DOIs	https://doi.org/10.1109/ICIS-COMSAR.2006.67
State	Published - 2006
Event	5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, Software Architecture and Reuse, COMSAR 2006 - Honolulu, HI, United States Duration: Jul 10 2006 → Jul 12 2006

Publication series

Name	Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006
Volume	2006

Conference

Conference	5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, Software Architecture and Reuse, COMSAR 2006
Country/Territory	United States
City	Honolulu, HI
Period	07/10/06 → 07/12/06

Keywords

Cayley graphs
Computer architecture
Matrix multiplication

Access to Document

10.1109/ICIS-COMSAR.2006.67

Cite this

Peng, M., Ramani, S., Tang, K. W., & Yavuz Oruç, A. (2006). Performance of PGA (Programmable Graph Architecture) for matrix multiplications. In Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S (pp. 361-367). Article 1652018 (Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006; Vol. 2006). https://doi.org/10.1109/ICIS-COMSAR.2006.67

Peng, Muling ; Ramani, Sreekanth ; Tang, K. Wendy et al. / Performance of PGA (Programmable Graph Architecture) for matrix multiplications. Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S. 2006. pp. 361-367 (Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006).

@inproceedings{1844a3402d5d4963824ed826bfc034d4,

title = "Performance of PGA (Programmable Graph Architecture) for matrix multiplications",

abstract = "Matrix multiplication is one of the primary operations in linear algebra and is useful in a wide spectrum of multimedia applications including signal and image processing. In this paper we review a novel computer architecture for matrix multiplications. This novel architecture is based on graph theory, hence the name Programmable Graph Architecture. We discuss the performance of PGA via time and space complexity and time-processor product for parallel implementation. We further compare the PGA performance with Cannon's algorithm and show that the PGA architecture outperforms the existing method for matrix with any dimension.",

keywords = "Cayley graphs, Computer architecture, Matrix multiplication",

author = "Muling Peng and Sreekanth Ramani and Tang, \{K. Wendy\} and \{Yavuz Oru{\c c}\}, A.",

year = "2006",

doi = "10.1109/ICIS-COMSAR.2006.67",

language = "English",

isbn = "0769526136",

series = "Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006",

pages = "361--367",

booktitle = "Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S",

note = "5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, Software Architecture and Reuse, COMSAR 2006 ; Conference date: 10-07-2006 Through 12-07-2006",

}

Peng, M, Ramani, S, Tang, KW & Yavuz Oruç, A 2006, Performance of PGA (Programmable Graph Architecture) for matrix multiplications. in Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S., 1652018, Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006, vol. 2006, pp. 361-367, 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, Software Architecture and Reuse, COMSAR 2006, Honolulu, HI, United States, 07/10/06. https://doi.org/10.1109/ICIS-COMSAR.2006.67

Performance of PGA (Programmable Graph Architecture) for matrix multiplications. / Peng, Muling; Ramani, Sreekanth; Tang, K. Wendy et al.
Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S. 2006. p. 361-367 1652018 (Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006; Vol. 2006).

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

TY - GEN

T1 - Performance of PGA (Programmable Graph Architecture) for matrix multiplications

AU - Peng, Muling

AU - Ramani, Sreekanth

AU - Tang, K. Wendy

AU - Yavuz Oruç, A.

PY - 2006

Y1 - 2006

N2 - Matrix multiplication is one of the primary operations in linear algebra and is useful in a wide spectrum of multimedia applications including signal and image processing. In this paper we review a novel computer architecture for matrix multiplications. This novel architecture is based on graph theory, hence the name Programmable Graph Architecture. We discuss the performance of PGA via time and space complexity and time-processor product for parallel implementation. We further compare the PGA performance with Cannon's algorithm and show that the PGA architecture outperforms the existing method for matrix with any dimension.

AB - Matrix multiplication is one of the primary operations in linear algebra and is useful in a wide spectrum of multimedia applications including signal and image processing. In this paper we review a novel computer architecture for matrix multiplications. This novel architecture is based on graph theory, hence the name Programmable Graph Architecture. We discuss the performance of PGA via time and space complexity and time-processor product for parallel implementation. We further compare the PGA performance with Cannon's algorithm and show that the PGA architecture outperforms the existing method for matrix with any dimension.

KW - Cayley graphs

KW - Computer architecture

KW - Matrix multiplication

UR - https://www.scopus.com/pages/publications/33947691255

U2 - 10.1109/ICIS-COMSAR.2006.67

DO - 10.1109/ICIS-COMSAR.2006.67

M3 - Conference contribution

AN - SCOPUS:33947691255

SN - 0769526136

SN - 9780769526133

T3 - Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006

SP - 361

EP - 367

BT - Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S

T2 - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, Software Architecture and Reuse, COMSAR 2006

Y2 - 10 July 2006 through 12 July 2006

ER -

Peng M, Ramani S, Tang KW, Yavuz Oruç A. Performance of PGA (Programmable Graph Architecture) for matrix multiplications. In Proceedings - 5th IEEE/ACIS International Conference on Computer and Information Science, ICIS 2006. In conjunction with 1st IEEE/ACIS International Workshop on Component-Based Software Engineering, S. 2006. p. 361-367. 1652018. (Proceedings - 5th IEEE/ACIS Int. Conf. on Comput. and Info. Sci., ICIS 2006. In conjunction with 1st IEEE/ACIS, Int. Workshop Component-Based Software Eng., Softw. Archi. and Reuse, COMSAR 2006). doi: 10.1109/ICIS-COMSAR.2006.67

Performance of PGA (Programmable Graph Architecture) for matrix multiplications

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this