An Abstraction for Distributed Stencil Computations Using Charm++

Aditya Bhosale; Zane Fink; Laxmikant Kale

doi:10.1007/978-3-031-61763-8_12

An Abstraction for Distributed Stencil Computations Using Charm++

Aditya Bhosale, Zane Fink, Laxmikant Kale

Siebel School of Computing and Data Science

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

Abstract

Python has emerged as a popular programming language for scientific computing in recent years, thanks to libraries like Numpy and SciPy. Numpy, in particular, is widely utilized for prototyping numerical solvers using methods such as finite difference, finite volume, and multigrid. However, Numpy’s performance is confined to a single node, compelling programmers to resort to a lower-level language for running large-scale simulations. In this paper, we introduce CharmStencil, a high-level abstraction featuring a Numpy-like Python frontend and a highly efficient Charm++ backend. Employing a client-server model, CharmStencil maintains productivity with tools like Jupyter notebooks on the frontend while utilizing a high-performance Charm++ library on the backend for computation. We demonstrate that CharmStencil achieves orders of magnitude better single-threaded performance compared to Numpy and can scale to thousands of CPU cores. Additionally, we showcase superior performance compared to cuNumeric and Numba, popular Python libraries for parallel array computations.

Original language	English (US)
Title of host publication	Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings
Editors	Patrick Diehl, Joseph Schuchart, Pedro Valero-Lara, George Bosilca
Publisher	Springer
Pages	123-134
Number of pages	12
ISBN (Print)	9783031617621
DOIs	https://doi.org/10.1007/978-3-031-61763-8_12
State	Published - 2024
Event	2nd International Workshop on Asynchronous Many-Task Systems and Applications, WAMTA 2024 - Knoxville, United States Duration: Feb 14 2024 → Feb 16 2024

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14626 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	2nd International Workshop on Asynchronous Many-Task Systems and Applications, WAMTA 2024
Country/Territory	United States
City	Knoxville
Period	2/14/24 → 2/16/24

Keywords

Charm++
Distributed
Numpy
Python
Stencil

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Online availability

10.1007/978-3-031-61763-8_12

Library availability

Discover UIUC Full Text

Cite this

Bhosale, A., Fink, Z., & Kale, L. (2024). An Abstraction for Distributed Stencil Computations Using Charm++. In P. Diehl, J. Schuchart, P. Valero-Lara, & G. Bosilca (Eds.), Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings (pp. 123-134). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14626 LNCS). Springer. https://doi.org/10.1007/978-3-031-61763-8_12

An Abstraction for Distributed Stencil Computations Using Charm++. / Bhosale, Aditya; Fink, Zane; Kale, Laxmikant.
Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings. ed. / Patrick Diehl; Joseph Schuchart; Pedro Valero-Lara; George Bosilca. Springer, 2024. p. 123-134 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14626 LNCS).

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

Bhosale, A, Fink, Z & Kale, L 2024, An Abstraction for Distributed Stencil Computations Using Charm++. in P Diehl, J Schuchart, P Valero-Lara & G Bosilca (eds), Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14626 LNCS, Springer, pp. 123-134, 2nd International Workshop on Asynchronous Many-Task Systems and Applications, WAMTA 2024, Knoxville, United States, 2/14/24. https://doi.org/10.1007/978-3-031-61763-8_12

Bhosale A, Fink Z, Kale L. An Abstraction for Distributed Stencil Computations Using Charm++. In Diehl P, Schuchart J, Valero-Lara P, Bosilca G, editors, Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings. Springer. 2024. p. 123-134. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-61763-8_12

Bhosale, Aditya ; Fink, Zane ; Kale, Laxmikant. / An Abstraction for Distributed Stencil Computations Using Charm++. Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings. editor / Patrick Diehl ; Joseph Schuchart ; Pedro Valero-Lara ; George Bosilca. Springer, 2024. pp. 123-134 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{d821a68aea89477bbc9a1a5e15156133,

title = "An Abstraction for Distributed Stencil Computations Using Charm++",

abstract = "Python has emerged as a popular programming language for scientific computing in recent years, thanks to libraries like Numpy and SciPy. Numpy, in particular, is widely utilized for prototyping numerical solvers using methods such as finite difference, finite volume, and multigrid. However, Numpy{\textquoteright}s performance is confined to a single node, compelling programmers to resort to a lower-level language for running large-scale simulations. In this paper, we introduce CharmStencil, a high-level abstraction featuring a Numpy-like Python frontend and a highly efficient Charm++ backend. Employing a client-server model, CharmStencil maintains productivity with tools like Jupyter notebooks on the frontend while utilizing a high-performance Charm++ library on the backend for computation. We demonstrate that CharmStencil achieves orders of magnitude better single-threaded performance compared to Numpy and can scale to thousands of CPU cores. Additionally, we showcase superior performance compared to cuNumeric and Numba, popular Python libraries for parallel array computations.",

keywords = "Charm++, Distributed, Numpy, Python, Stencil",

author = "Aditya Bhosale and Zane Fink and Laxmikant Kale",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; 2nd International Workshop on Asynchronous Many-Task Systems and Applications, WAMTA 2024 ; Conference date: 14-02-2024 Through 16-02-2024",

year = "2024",

doi = "10.1007/978-3-031-61763-8_12",

language = "English (US)",

isbn = "9783031617621",

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

publisher = "Springer",

pages = "123--134",

editor = "Patrick Diehl and Joseph Schuchart and Pedro Valero-Lara and George Bosilca",

booktitle = "Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings",

address = "Germany",

}

TY - GEN

T1 - An Abstraction for Distributed Stencil Computations Using Charm++

AU - Bhosale, Aditya

AU - Fink, Zane

AU - Kale, Laxmikant

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

PY - 2024

Y1 - 2024

N2 - Python has emerged as a popular programming language for scientific computing in recent years, thanks to libraries like Numpy and SciPy. Numpy, in particular, is widely utilized for prototyping numerical solvers using methods such as finite difference, finite volume, and multigrid. However, Numpy’s performance is confined to a single node, compelling programmers to resort to a lower-level language for running large-scale simulations. In this paper, we introduce CharmStencil, a high-level abstraction featuring a Numpy-like Python frontend and a highly efficient Charm++ backend. Employing a client-server model, CharmStencil maintains productivity with tools like Jupyter notebooks on the frontend while utilizing a high-performance Charm++ library on the backend for computation. We demonstrate that CharmStencil achieves orders of magnitude better single-threaded performance compared to Numpy and can scale to thousands of CPU cores. Additionally, we showcase superior performance compared to cuNumeric and Numba, popular Python libraries for parallel array computations.

AB - Python has emerged as a popular programming language for scientific computing in recent years, thanks to libraries like Numpy and SciPy. Numpy, in particular, is widely utilized for prototyping numerical solvers using methods such as finite difference, finite volume, and multigrid. However, Numpy’s performance is confined to a single node, compelling programmers to resort to a lower-level language for running large-scale simulations. In this paper, we introduce CharmStencil, a high-level abstraction featuring a Numpy-like Python frontend and a highly efficient Charm++ backend. Employing a client-server model, CharmStencil maintains productivity with tools like Jupyter notebooks on the frontend while utilizing a high-performance Charm++ library on the backend for computation. We demonstrate that CharmStencil achieves orders of magnitude better single-threaded performance compared to Numpy and can scale to thousands of CPU cores. Additionally, we showcase superior performance compared to cuNumeric and Numba, popular Python libraries for parallel array computations.

KW - Charm++

KW - Distributed

KW - Numpy

KW - Python

KW - Stencil

UR - http://www.scopus.com/inward/record.url?scp=85197218750&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85197218750&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-61763-8_12

DO - 10.1007/978-3-031-61763-8_12

M3 - Conference contribution

AN - SCOPUS:85197218750

SN - 9783031617621

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

SP - 123

EP - 134

BT - Asynchronous Many-Task Systems and Applications - 2nd International Workshop, WAMTA 2024, Proceedings

A2 - Diehl, Patrick

A2 - Schuchart, Joseph

A2 - Valero-Lara, Pedro

A2 - Bosilca, George

PB - Springer

T2 - 2nd International Workshop on Asynchronous Many-Task Systems and Applications, WAMTA 2024

Y2 - 14 February 2024 through 16 February 2024

ER -