IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library

Narasinga Rao Miniskar; Mohammad Alaul Haque Monil; Pedro Valero-Lara; Frank Liu; Jeffrey S. Vetter

doi:10.1109/HiPC56025.2022.00042

IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library

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

10 Scopus citations

Abstract

This paper presents IRIS-BLAS, a novel heterogeneous and performance portable BLAS library. IRIS-BLAS is built on top of the IRIS runtime and multiple vendor and open-source BLAS libraries. It can transparently use all the architectures/devices available in a heterogeneous system, using the appropriate BLAS library based on the task mapping at run time. Thus, IRIS-BLAS is portable across a broad spectrum of architectures and BLAS libraries, alleviating the worry of application developers about modifying the application source code. Even though the emphasis is on portability, IRIS-BLAS provides competitive or even better performance than other state-of-the-art references. Moreover, IRIS-BLAS offers new features such as efficiently using extremely heterogeneous systems composed of multiple GPUs from different hardware vendors.

Original language	English
Title of host publication	Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	256-261
Number of pages	6
ISBN (Electronic)	9781665494236
DOIs	https://doi.org/10.1109/HiPC56025.2022.00042
State	Published - 2022
Event	29th Annual IEEE International Conference on High Performance Computing, Data, and Analytics, HiPC 2022 - Bangalore, India Duration: Dec 18 2022 → Dec 21 2022

Publication series

Name	Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022

Conference

Conference	29th Annual IEEE International Conference on High Performance Computing, Data, and Analytics, HiPC 2022
Country/Territory	India
City	Bangalore
Period	12/18/22 → 12/21/22

Funding

Notice: This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725with the U.S. Department of Energy. The publisher, by accepting the article for publication, acknowledgesthat the U.S. Government retains a non-exclusive, paid up, irrevocable, world-wide license to

Keywords

BLAS
Heterogeneity
IRIS
Performance Portable
Tasking

Access to Document

10.1109/HiPC56025.2022.00042

Cite this

Miniskar, N. R., Monil, M. A. H., Valero-Lara, P., Liu, F., & Vetter, J. S. (2022). IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library. In Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022 (pp. 256-261). (Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/HiPC56025.2022.00042

Miniskar, Narasinga Rao ; Monil, Mohammad Alaul Haque ; Valero-Lara, Pedro et al. / IRIS-BLAS : Towards a Performance Portable and Heterogeneous BLAS Library. Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022. Institute of Electrical and Electronics Engineers Inc., 2022. pp. 256-261 (Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022).

@inproceedings{374b7886b1f24ec0a94e5d53ef52898e,

title = "IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library",

abstract = "This paper presents IRIS-BLAS, a novel heterogeneous and performance portable BLAS library. IRIS-BLAS is built on top of the IRIS runtime and multiple vendor and open-source BLAS libraries. It can transparently use all the architectures/devices available in a heterogeneous system, using the appropriate BLAS library based on the task mapping at run time. Thus, IRIS-BLAS is portable across a broad spectrum of architectures and BLAS libraries, alleviating the worry of application developers about modifying the application source code. Even though the emphasis is on portability, IRIS-BLAS provides competitive or even better performance than other state-of-the-art references. Moreover, IRIS-BLAS offers new features such as efficiently using extremely heterogeneous systems composed of multiple GPUs from different hardware vendors.",

keywords = "BLAS, Heterogeneity, IRIS, Performance Portable, Tasking",

author = "Miniskar, \{Narasinga Rao\} and Monil, \{Mohammad Alaul Haque\} and Pedro Valero-Lara and Frank Liu and Vetter, \{Jeffrey S.\}",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 29th Annual IEEE International Conference on High Performance Computing, Data, and Analytics, HiPC 2022 ; Conference date: 18-12-2022 Through 21-12-2022",

year = "2022",

doi = "10.1109/HiPC56025.2022.00042",

language = "English",

series = "Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "256--261",

booktitle = "Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022",

}

Miniskar, NR , Monil, MAH , Valero-Lara, P, Liu, F & Vetter, JS 2022, IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library. in Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022. Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022, Institute of Electrical and Electronics Engineers Inc., pp. 256-261, 29th Annual IEEE International Conference on High Performance Computing, Data, and Analytics, HiPC 2022, Bangalore, India, 12/18/22. https://doi.org/10.1109/HiPC56025.2022.00042

IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library. / Miniskar, Narasinga Rao ; Monil, Mohammad Alaul Haque ; Valero-Lara, Pedro et al.
Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022. Institute of Electrical and Electronics Engineers Inc., 2022. p. 256-261 (Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022).

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

TY - GEN

T1 - IRIS-BLAS

T2 - 29th Annual IEEE International Conference on High Performance Computing, Data, and Analytics, HiPC 2022

AU - Miniskar, Narasinga Rao

AU - Monil, Mohammad Alaul Haque

AU - Valero-Lara, Pedro

AU - Liu, Frank

AU - Vetter, Jeffrey S.

PY - 2022

Y1 - 2022

N2 - This paper presents IRIS-BLAS, a novel heterogeneous and performance portable BLAS library. IRIS-BLAS is built on top of the IRIS runtime and multiple vendor and open-source BLAS libraries. It can transparently use all the architectures/devices available in a heterogeneous system, using the appropriate BLAS library based on the task mapping at run time. Thus, IRIS-BLAS is portable across a broad spectrum of architectures and BLAS libraries, alleviating the worry of application developers about modifying the application source code. Even though the emphasis is on portability, IRIS-BLAS provides competitive or even better performance than other state-of-the-art references. Moreover, IRIS-BLAS offers new features such as efficiently using extremely heterogeneous systems composed of multiple GPUs from different hardware vendors.

AB - This paper presents IRIS-BLAS, a novel heterogeneous and performance portable BLAS library. IRIS-BLAS is built on top of the IRIS runtime and multiple vendor and open-source BLAS libraries. It can transparently use all the architectures/devices available in a heterogeneous system, using the appropriate BLAS library based on the task mapping at run time. Thus, IRIS-BLAS is portable across a broad spectrum of architectures and BLAS libraries, alleviating the worry of application developers about modifying the application source code. Even though the emphasis is on portability, IRIS-BLAS provides competitive or even better performance than other state-of-the-art references. Moreover, IRIS-BLAS offers new features such as efficiently using extremely heterogeneous systems composed of multiple GPUs from different hardware vendors.

KW - BLAS

KW - Heterogeneity

KW - IRIS

KW - Performance Portable

KW - Tasking

UR - https://www.scopus.com/pages/publications/85148236269

U2 - 10.1109/HiPC56025.2022.00042

DO - 10.1109/HiPC56025.2022.00042

M3 - Conference contribution

AN - SCOPUS:85148236269

T3 - Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022

SP - 256

EP - 261

BT - Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 18 December 2022 through 21 December 2022

ER -

Miniskar NR , Monil MAH , Valero-Lara P, Liu F, Vetter JS. IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library. In Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022. Institute of Electrical and Electronics Engineers Inc. 2022. p. 256-261. (Proceedings - 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics, HiPC 2022). doi: 10.1109/HiPC56025.2022.00042

IRIS-BLAS: Towards a Performance Portable and Heterogeneous BLAS Library

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this