MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime

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

doi:10.1145/3624062.3624184

MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime

Mohammad Alaul Haque Monil, Narasinga Rao Miniskar, Keita Teranishi, Jeffrey S. Vetter, Pedro Valero-Lara

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

4 Scopus citations

Abstract

Vendor libraries are tuned for a specific architecture and are not portable to others. Moreover, they lack support for heterogeneity and multi-device orchestration, which is required for efficient use of contemporary HPC and cloud resources. To address these challenges, we introduce MatRIS - a multilevel math library abstraction for scalable and performance-portable sparse/dense BLAS/LAPACK operations using IRIS runtime. The MatRIS-IRIS co-design introduces three levels of abstraction to make the implementation completely architecture agnostic and provide highly productive programming. We demonstrate that MatRIS is portable without any change in source code and can fully utilize multi-device heterogeneous systems by achieving high performance and scalability on Summit, Frontier, and a CADES cloud node equipped with four NVIDIA A100 GPUs and four AMD MI100 GPUs. A detailed performance study is presented in which MatRIS demonstrates multi-device scalability. When compared, MatRIS provides competitive and even better performance than libraries from vendors and other third parties.

Original language	English
Title of host publication	Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023
Publisher	Association for Computing Machinery
Pages	1081-1092
Number of pages	12
ISBN (Electronic)	9798400707858
DOIs	https://doi.org/10.1145/3624062.3624184
State	Published - Nov 12 2023
Event	2023 International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023 - Denver, United States Duration: Nov 12 2023 → Nov 17 2023

Publication series

Name	ACM International Conference Proceeding Series

Conference

Conference	2023 International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023
Country/Territory	United States
City	Denver
Period	11/12/23 → 11/17/23

Funding

This research used resources of the Oak Ridge Leadership Computing Facility at ORNL, which is supported by the US Department of Energy’s (DOE’s) Office of Science under Contract No. DE-AC05-00OR22725.This work is funded, in part, by Bluestone, a X-Stack project in the DOE Advanced Scientific Computing Office with program manager Hal Finkel. This manuscript has been authored by UT-Battelle LLC under contract no. DE-AC05-00OR22725 with the US Department of Energy. The publisher, by accepting the article for publication, acknowledges that the US government retains a non-exclusive, paid up, irrevocable, world-wide license to publish or reproduce the published form of the manuscript, or allow others to do so, for US government purposes. The DOE will provide public access to these results in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Keywords

BLAS
GPUs
LAPACK
LU factorization
Performance portability
extreme heterogeneity
programming productivity
scalability

Access to Document

10.1145/3624062.3624184

Cite this

Monil, M. A. H., Miniskar, N. R., Teranishi, K., Vetter, J. S., & Valero-Lara, P. (2023). MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime. In Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023 (pp. 1081-1092). (ACM International Conference Proceeding Series). Association for Computing Machinery. https://doi.org/10.1145/3624062.3624184

Monil, Mohammad Alaul Haque ; Miniskar, Narasinga Rao ; Teranishi, Keita et al. / MatRIS : Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime. Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023. Association for Computing Machinery, 2023. pp. 1081-1092 (ACM International Conference Proceeding Series).

@inproceedings{8a4163a174bd4bb1a911a55a7d946f5c,

title = "MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime",

abstract = "Vendor libraries are tuned for a specific architecture and are not portable to others. Moreover, they lack support for heterogeneity and multi-device orchestration, which is required for efficient use of contemporary HPC and cloud resources. To address these challenges, we introduce MatRIS - a multilevel math library abstraction for scalable and performance-portable sparse/dense BLAS/LAPACK operations using IRIS runtime. The MatRIS-IRIS co-design introduces three levels of abstraction to make the implementation completely architecture agnostic and provide highly productive programming. We demonstrate that MatRIS is portable without any change in source code and can fully utilize multi-device heterogeneous systems by achieving high performance and scalability on Summit, Frontier, and a CADES cloud node equipped with four NVIDIA A100 GPUs and four AMD MI100 GPUs. A detailed performance study is presented in which MatRIS demonstrates multi-device scalability. When compared, MatRIS provides competitive and even better performance than libraries from vendors and other third parties.",

keywords = "BLAS, GPUs, LAPACK, LU factorization, Performance portability, extreme heterogeneity, programming productivity, scalability",

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

note = "Publisher Copyright: {\textcopyright} 2023 ACM.; 2023 International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023 ; Conference date: 12-11-2023 Through 17-11-2023",

year = "2023",

month = nov,

day = "12",

doi = "10.1145/3624062.3624184",

language = "English",

series = "ACM International Conference Proceeding Series",

publisher = "Association for Computing Machinery",

pages = "1081--1092",

booktitle = "Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023",

}

Monil, MAH , Miniskar, NR, Teranishi, K, Vetter, JS & Valero-Lara, P 2023, MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime. in Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023. ACM International Conference Proceeding Series, Association for Computing Machinery, pp. 1081-1092, 2023 International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023, Denver, United States, 11/12/23. https://doi.org/10.1145/3624062.3624184

MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime. / Monil, Mohammad Alaul Haque ; Miniskar, Narasinga Rao; Teranishi, Keita et al.
Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023. Association for Computing Machinery, 2023. p. 1081-1092 (ACM International Conference Proceeding Series).

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

TY - GEN

T1 - MatRIS

T2 - 2023 International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023

AU - Monil, Mohammad Alaul Haque

AU - Miniskar, Narasinga Rao

AU - Teranishi, Keita

AU - Vetter, Jeffrey S.

AU - Valero-Lara, Pedro

PY - 2023/11/12

Y1 - 2023/11/12

N2 - Vendor libraries are tuned for a specific architecture and are not portable to others. Moreover, they lack support for heterogeneity and multi-device orchestration, which is required for efficient use of contemporary HPC and cloud resources. To address these challenges, we introduce MatRIS - a multilevel math library abstraction for scalable and performance-portable sparse/dense BLAS/LAPACK operations using IRIS runtime. The MatRIS-IRIS co-design introduces three levels of abstraction to make the implementation completely architecture agnostic and provide highly productive programming. We demonstrate that MatRIS is portable without any change in source code and can fully utilize multi-device heterogeneous systems by achieving high performance and scalability on Summit, Frontier, and a CADES cloud node equipped with four NVIDIA A100 GPUs and four AMD MI100 GPUs. A detailed performance study is presented in which MatRIS demonstrates multi-device scalability. When compared, MatRIS provides competitive and even better performance than libraries from vendors and other third parties.

AB - Vendor libraries are tuned for a specific architecture and are not portable to others. Moreover, they lack support for heterogeneity and multi-device orchestration, which is required for efficient use of contemporary HPC and cloud resources. To address these challenges, we introduce MatRIS - a multilevel math library abstraction for scalable and performance-portable sparse/dense BLAS/LAPACK operations using IRIS runtime. The MatRIS-IRIS co-design introduces three levels of abstraction to make the implementation completely architecture agnostic and provide highly productive programming. We demonstrate that MatRIS is portable without any change in source code and can fully utilize multi-device heterogeneous systems by achieving high performance and scalability on Summit, Frontier, and a CADES cloud node equipped with four NVIDIA A100 GPUs and four AMD MI100 GPUs. A detailed performance study is presented in which MatRIS demonstrates multi-device scalability. When compared, MatRIS provides competitive and even better performance than libraries from vendors and other third parties.

KW - BLAS

KW - GPUs

KW - LAPACK

KW - LU factorization

KW - Performance portability

KW - extreme heterogeneity

KW - programming productivity

KW - scalability

UR - http://www.scopus.com/inward/record.url?scp=85178119782&partnerID=8YFLogxK

U2 - 10.1145/3624062.3624184

DO - 10.1145/3624062.3624184

M3 - Conference contribution

AN - SCOPUS:85178119782

T3 - ACM International Conference Proceeding Series

SP - 1081

EP - 1092

BT - Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023

PB - Association for Computing Machinery

Y2 - 12 November 2023 through 17 November 2023

ER -

Monil MAH , Miniskar NR, Teranishi K, Vetter JS , Valero-Lara P. MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime. In Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023. Association for Computing Machinery. 2023. p. 1081-1092. (ACM International Conference Proceeding Series). doi: 10.1145/3624062.3624184

MatRIS: Multi-level Math Library Abstraction for Heterogeneity and Performance Portability using IRIS Runtime

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this