Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs

Isha Aggarwal; Aditya Kashi; Pratik Nayak; Cody J. Balos; Carol S. Woodward; Hartwig Anzt

doi:10.1109/ScalA54577.2021.00010

Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs

Isha Aggarwal, Aditya Kashi, Pratik Nayak, Cody J. Balos, Carol S. Woodward, Hartwig Anzt

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

7 Scopus citations

Abstract

This paper presents batched iterative solvers for GPU architectures. We elaborate on the design of the batched functionality aiming for optimal performance while still giving the user some flexibility in terms of choosing a sparse matrix format, a preconditioner optimized for the distinct items of the batch, and an application-specific stopping criterion that is evaluated for each problem in the batch, individually. Performance results for benchmark problems coming from PeleLM simulations reveal the potential of the batched iterative solvers for computational chemistry simulations, and their advantage compared to the current vendor-provided batched solutions.

Original language	English
Title of host publication	Proceedings of ScalA 2021
Subtitle of host publication	12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	35-43
Number of pages	9
ISBN (Electronic)	9781665411288
DOIs	https://doi.org/10.1109/ScalA54577.2021.00010
State	Published - 2021
Externally published	Yes
Event	12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, ScalA 2021 - St. Louis, United States Duration: Nov 19 2021 → …

Publication series

Name	Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis

Conference

Conference	12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, ScalA 2021
Country/Territory	United States
City	St. Louis
Period	11/19/21 → …

Funding

This research was supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration. Isha Aggarwal, Aditya Kashi, Pratik Nayak, and Hartwig Anzt were also supported by the “Impuls und Vernetzungsfond” of the Helmholtz Association under grant VH-NG-1241. Prepared by LLNL under Contract DE-AC52-07NA27344. LLNL-PROC-826165.

Funders	Funder number
U.S. Department of Energy
National Nuclear Security Administration
Helmholtz Association	VH-NG-1241, DE-AC52-07NA27344, LLNL-PROC-826165

Keywords

GPU
Ginkgo
Sparse linear systems
batched solvers

Access to Document

10.1109/ScalA54577.2021.00010

Cite this

Aggarwal, I., Kashi, A., Nayak, P., Balos, C. J., Woodward, C. S., & Anzt, H. (2021). Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs. In Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis (pp. 35-43). (Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ScalA54577.2021.00010

Aggarwal, Isha ; Kashi, Aditya ; Nayak, Pratik et al. / Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs. Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 35-43 (Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis).

@inproceedings{7ab21128756c4471b27dfa996e8292eb,

title = "Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs",

abstract = "This paper presents batched iterative solvers for GPU architectures. We elaborate on the design of the batched functionality aiming for optimal performance while still giving the user some flexibility in terms of choosing a sparse matrix format, a preconditioner optimized for the distinct items of the batch, and an application-specific stopping criterion that is evaluated for each problem in the batch, individually. Performance results for benchmark problems coming from PeleLM simulations reveal the potential of the batched iterative solvers for computational chemistry simulations, and their advantage compared to the current vendor-provided batched solutions.",

keywords = "GPU, Ginkgo, Sparse linear systems, batched solvers",

author = "Isha Aggarwal and Aditya Kashi and Pratik Nayak and Balos, {Cody J.} and Woodward, {Carol S.} and Hartwig Anzt",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, ScalA 2021 ; Conference date: 19-11-2021",

year = "2021",

doi = "10.1109/ScalA54577.2021.00010",

language = "English",

series = "Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis",

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

pages = "35--43",

booktitle = "Proceedings of ScalA 2021",

}

Aggarwal, I, Kashi, A, Nayak, P, Balos, CJ, Woodward, CS & Anzt, H 2021, Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs. in Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis. Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis, Institute of Electrical and Electronics Engineers Inc., pp. 35-43, 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, ScalA 2021, St. Louis, United States, 11/19/21. https://doi.org/10.1109/ScalA54577.2021.00010

Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs. / Aggarwal, Isha; Kashi, Aditya; Nayak, Pratik et al.
Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc., 2021. p. 35-43 (Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis).

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

TY - GEN

T1 - Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs

AU - Aggarwal, Isha

AU - Kashi, Aditya

AU - Nayak, Pratik

AU - Balos, Cody J.

AU - Woodward, Carol S.

AU - Anzt, Hartwig

PY - 2021

Y1 - 2021

N2 - This paper presents batched iterative solvers for GPU architectures. We elaborate on the design of the batched functionality aiming for optimal performance while still giving the user some flexibility in terms of choosing a sparse matrix format, a preconditioner optimized for the distinct items of the batch, and an application-specific stopping criterion that is evaluated for each problem in the batch, individually. Performance results for benchmark problems coming from PeleLM simulations reveal the potential of the batched iterative solvers for computational chemistry simulations, and their advantage compared to the current vendor-provided batched solutions.

AB - This paper presents batched iterative solvers for GPU architectures. We elaborate on the design of the batched functionality aiming for optimal performance while still giving the user some flexibility in terms of choosing a sparse matrix format, a preconditioner optimized for the distinct items of the batch, and an application-specific stopping criterion that is evaluated for each problem in the batch, individually. Performance results for benchmark problems coming from PeleLM simulations reveal the potential of the batched iterative solvers for computational chemistry simulations, and their advantage compared to the current vendor-provided batched solutions.

KW - GPU

KW - Ginkgo

KW - Sparse linear systems

KW - batched solvers

UR - http://www.scopus.com/inward/record.url?scp=85124661523&partnerID=8YFLogxK

U2 - 10.1109/ScalA54577.2021.00010

DO - 10.1109/ScalA54577.2021.00010

M3 - Conference contribution

AN - SCOPUS:85124661523

T3 - Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis

SP - 35

EP - 43

BT - Proceedings of ScalA 2021

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, ScalA 2021

Y2 - 19 November 2021

ER -

Aggarwal I, Kashi A, Nayak P, Balos CJ, Woodward CS, Anzt H. Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs. In Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc. 2021. p. 35-43. (Proceedings of ScalA 2021: 12th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, Held in conjunction with SC 2021: The International Conference for High Performance Computing, Networking, Storage and Analysis). doi: 10.1109/ScalA54577.2021.00010

Batched Sparse Iterative Solvers for Computational Chemistry Simulations on GPUs

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this