Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations

Takahiro Katagiri; Takao Sakurai; Mitsuyoshi Igai; Satoshi Ohshima; Hisayasu Kuroda; Ken Naono; Kengo Nakajima

doi:10.1007/978-3-642-38718-0_24

Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations

Takahiro Katagiri, Takao Sakurai, Mitsuyoshi Igai, Satoshi Ohshima, Hisayasu Kuroda, Ken Naono, Kengo Nakajima

研究成果: 書籍/レポートタイプへの寄稿 › 会議への寄与

2 被引用数 (Scopus)

抄録

In this paper, we propose "control formats" to obtain better thread performance of sparse matrix-vector multiplication (SpMV) for unsymmetric and symmetric matrices. By using the control formats, we established the following maximum speedups of SpMV in 16-thread execution on one node of the T2K Open Supercomputer: (1) 7.14x for an unsymmetric matrix by using the proposed Branchless Segmented Scan compared to the original Segmented Scan method; (2) 12.7x for a symmetric matrix by using the proposed Zero-element Computation-free method compared to a simple SpMV implementation.

本文言語	英語
ホスト出版物のタイトル	High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers
ページ	236-248
ページ数	13
DOI	https://doi.org/10.1007/978-3-642-38718-0_24
出版ステータス	出版済み - 9月 5 2013
外部発表	はい
イベント	10th International Conference on High Performance Computing for Computational Science, VECPAR 2012 - Kobe, 日本継続期間: 7月 17 2012 → 7月 20 2012

出版物シリーズ

名前	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
巻	7851 LNCS
ISSN（印刷版）	0302-9743
ISSN（電子版）	1611-3349

その他

その他	10th International Conference on High Performance Computing for Computational Science, VECPAR 2012
国/地域	日本
City	Kobe
Period	7/17/12 → 7/20/12

!!!All Science Journal Classification (ASJC) codes

理論的コンピュータサイエンス
コンピュータサイエンス（全般）

文献へのアクセス

10.1007/978-3-642-38718-0_24

他のファイルとリンク

引用スタイル

Katagiri, T., Sakurai, T., Igai, M., Ohshima, S., Kuroda, H., Naono, K., & Nakajima, K. (2013). Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations. In High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers (pp. 236-248). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7851 LNCS). https://doi.org/10.1007/978-3-642-38718-0_24

Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations. / Katagiri, Takahiro; Sakurai, Takao; Igai, Mitsuyoshi その他.
High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers. 2013. p. 236-248 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7851 LNCS).

研究成果: 書籍/レポートタイプへの寄稿 › 会議への寄与

Katagiri, T, Sakurai, T, Igai, M, Ohshima, S, Kuroda, H, Naono, K & Nakajima, K 2013, Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations. in High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7851 LNCS, pp. 236-248, 10th International Conference on High Performance Computing for Computational Science, VECPAR 2012, Kobe, 日本, 7/17/12. https://doi.org/10.1007/978-3-642-38718-0_24

Katagiri T, Sakurai T, Igai M, Ohshima S, Kuroda H, Naono K その他. Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations. In High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers. 2013. p. 236-248. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-38718-0_24

Katagiri, Takahiro ; Sakurai, Takao ; Igai, Mitsuyoshi その他. / Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations. High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers. 2013. pp. 236-248 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{cbbc3c4abf374a8ebc02aa36320a1c87,

title = "Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations",

abstract = "In this paper, we propose {"}control formats{"} to obtain better thread performance of sparse matrix-vector multiplication (SpMV) for unsymmetric and symmetric matrices. By using the control formats, we established the following maximum speedups of SpMV in 16-thread execution on one node of the T2K Open Supercomputer: (1) 7.14x for an unsymmetric matrix by using the proposed Branchless Segmented Scan compared to the original Segmented Scan method; (2) 12.7x for a symmetric matrix by using the proposed Zero-element Computation-free method compared to a simple SpMV implementation.",

author = "Takahiro Katagiri and Takao Sakurai and Mitsuyoshi Igai and Satoshi Ohshima and Hisayasu Kuroda and Ken Naono and Kengo Nakajima",

year = "2013",

month = sep,

day = "5",

doi = "10.1007/978-3-642-38718-0_24",

language = "English",

isbn = "9783642387173",

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

pages = "236--248",

booktitle = "High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers",

note = "10th International Conference on High Performance Computing for Computational Science, VECPAR 2012 ; Conference date: 17-07-2012 Through 20-07-2012",

}

TY - GEN

T1 - Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations

AU - Katagiri, Takahiro

AU - Sakurai, Takao

AU - Igai, Mitsuyoshi

AU - Ohshima, Satoshi

AU - Kuroda, Hisayasu

AU - Naono, Ken

AU - Nakajima, Kengo

PY - 2013/9/5

Y1 - 2013/9/5

N2 - In this paper, we propose "control formats" to obtain better thread performance of sparse matrix-vector multiplication (SpMV) for unsymmetric and symmetric matrices. By using the control formats, we established the following maximum speedups of SpMV in 16-thread execution on one node of the T2K Open Supercomputer: (1) 7.14x for an unsymmetric matrix by using the proposed Branchless Segmented Scan compared to the original Segmented Scan method; (2) 12.7x for a symmetric matrix by using the proposed Zero-element Computation-free method compared to a simple SpMV implementation.

AB - In this paper, we propose "control formats" to obtain better thread performance of sparse matrix-vector multiplication (SpMV) for unsymmetric and symmetric matrices. By using the control formats, we established the following maximum speedups of SpMV in 16-thread execution on one node of the T2K Open Supercomputer: (1) 7.14x for an unsymmetric matrix by using the proposed Branchless Segmented Scan compared to the original Segmented Scan method; (2) 12.7x for a symmetric matrix by using the proposed Zero-element Computation-free method compared to a simple SpMV implementation.

UR - http://www.scopus.com/inward/record.url?scp=84883285608&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84883285608&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-38718-0_24

DO - 10.1007/978-3-642-38718-0_24

M3 - Conference contribution

AN - SCOPUS:84883285608

SN - 9783642387173

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

SP - 236

EP - 248

BT - High Performance Computing for Computational Science, VECPAR 2012 - 10th International Conference, Revised Selected Papers

T2 - 10th International Conference on High Performance Computing for Computational Science, VECPAR 2012

Y2 - 17 July 2012 through 20 July 2012

ER -

Control formats for unsymmetric and symmetric sparse matrix-vector multiplications on OpenMP implementations

抄録

出版物シリーズ

その他

!!!All Science Journal Classification (ASJC) codes

文献へのアクセス

他のファイルとリンク

フィンガープリント

引用スタイル