Sharing-Aware Data Mapping in Software Transactional Memory

Douglas Pereira Pasqualin; Matthias Diener; André Rauber Du Bois; Maurício Lima Pilla

doi:10.1007/978-3-031-04580-6_32

Sharing-Aware Data Mapping in Software Transactional Memory

Douglas Pereira Pasqualin, Matthias Diener, André Rauber Du Bois, Maurício Lima Pilla

National Center for Supercomputing Applications (NCSA)

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

Abstract

Software transactional memory (STM) is an abstraction used for thread synchronization that borrows the concept of transactions from databases. It is often easier to use than locks, proving a high-level abstraction for software developers. In current multicore architectures, data locality is an important aspect of STM performance. Sharing-aware mapping is a technique that aims to improve the performance of applications by mapping threads and data (in the form of memory pages) according to their memory access behavior. In prior work, we successfully used information gained from tracking STM variables and STM operations to perform an effective sharing-aware thread mapping. In this paper, we attempt to extend such a mechanism to perform data mapping. Although initial results using a synthetic application were encouraging, data mapping did not improve performance when using realistic workloads. Contrary to thread mapping, where only keeping track of STM operations is sufficient to perform an effective thread mapping, data mapping requires a global vision of memory page accesses of the application to be able to improve the performance, which STM runtimes can not provide.

Original language	English (US)
Title of host publication	Embedded Computer Systems
Subtitle of host publication	Architectures, Modeling, and Simulation - 21st International Conference, SAMOS 2021, Proceedings
Editors	Alex Orailoglu, Matthias Jung, Marc Reichenbach
Publisher	Springer
Pages	481-492
Number of pages	12
ISBN (Print)	9783031045790
DOIs	https://doi.org/10.1007/978-3-031-04580-6_32
State	Published - 2022
Event	21st International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2021 - Virtual, Online Duration: Jul 4 2021 → Jul 8 2021

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13227 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	21st International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2021
City	Virtual, Online
Period	7/4/21 → 7/8/21

Keywords

Data mapping
NUMA
Sharing-aware
Software transactional memory

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Online availability

10.1007/978-3-031-04580-6_32

Library availability

Discover UIUC Full Text

Cite this

Pasqualin, D. P., Diener, M., Du Bois, A. R., & Pilla, M. L. (2022). Sharing-Aware Data Mapping in Software Transactional Memory. In A. Orailoglu, M. Jung, & M. Reichenbach (Eds.), Embedded Computer Systems: Architectures, Modeling, and Simulation - 21st International Conference, SAMOS 2021, Proceedings (pp. 481-492). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13227 LNCS). Springer. https://doi.org/10.1007/978-3-031-04580-6_32

Sharing-Aware Data Mapping in Software Transactional Memory. / Pasqualin, Douglas Pereira; Diener, Matthias; Du Bois, André Rauber et al.
Embedded Computer Systems: Architectures, Modeling, and Simulation - 21st International Conference, SAMOS 2021, Proceedings. ed. / Alex Orailoglu; Matthias Jung; Marc Reichenbach. Springer, 2022. p. 481-492 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13227 LNCS).

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

Pasqualin, DP, Diener, M, Du Bois, AR & Pilla, ML 2022, Sharing-Aware Data Mapping in Software Transactional Memory. in A Orailoglu, M Jung & M Reichenbach (eds), Embedded Computer Systems: Architectures, Modeling, and Simulation - 21st International Conference, SAMOS 2021, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13227 LNCS, Springer, pp. 481-492, 21st International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2021, Virtual, Online, 7/4/21. https://doi.org/10.1007/978-3-031-04580-6_32

Pasqualin DP, Diener M, Du Bois AR, Pilla ML. Sharing-Aware Data Mapping in Software Transactional Memory. In Orailoglu A, Jung M, Reichenbach M, editors, Embedded Computer Systems: Architectures, Modeling, and Simulation - 21st International Conference, SAMOS 2021, Proceedings. Springer. 2022. p. 481-492. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-04580-6_32

Pasqualin, Douglas Pereira ; Diener, Matthias ; Du Bois, André Rauber et al. / Sharing-Aware Data Mapping in Software Transactional Memory. Embedded Computer Systems: Architectures, Modeling, and Simulation - 21st International Conference, SAMOS 2021, Proceedings. editor / Alex Orailoglu ; Matthias Jung ; Marc Reichenbach. Springer, 2022. pp. 481-492 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{424b7b52b7c943b59f4f2c28e7eb0470,

title = "Sharing-Aware Data Mapping in Software Transactional Memory",

abstract = "Software transactional memory (STM) is an abstraction used for thread synchronization that borrows the concept of transactions from databases. It is often easier to use than locks, proving a high-level abstraction for software developers. In current multicore architectures, data locality is an important aspect of STM performance. Sharing-aware mapping is a technique that aims to improve the performance of applications by mapping threads and data (in the form of memory pages) according to their memory access behavior. In prior work, we successfully used information gained from tracking STM variables and STM operations to perform an effective sharing-aware thread mapping. In this paper, we attempt to extend such a mechanism to perform data mapping. Although initial results using a synthetic application were encouraging, data mapping did not improve performance when using realistic workloads. Contrary to thread mapping, where only keeping track of STM operations is sufficient to perform an effective thread mapping, data mapping requires a global vision of memory page accesses of the application to be able to improve the performance, which STM runtimes can not provide.",

keywords = "Data mapping, NUMA, Sharing-aware, Software transactional memory",

author = "Pasqualin, {Douglas Pereira} and Matthias Diener and {Du Bois}, {Andr{\'e} Rauber} and Pilla, {Maur{\'i}cio Lima}",

note = "Publisher Copyright: {\textcopyright} 2022, Springer Nature Switzerland AG.; 21st International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2021 ; Conference date: 04-07-2021 Through 08-07-2021",

year = "2022",

doi = "10.1007/978-3-031-04580-6_32",

language = "English (US)",

isbn = "9783031045790",

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

publisher = "Springer",

pages = "481--492",

editor = "Alex Orailoglu and Matthias Jung and Marc Reichenbach",

booktitle = "Embedded Computer Systems",

address = "Germany",

}

TY - GEN

T1 - Sharing-Aware Data Mapping in Software Transactional Memory

AU - Pasqualin, Douglas Pereira

AU - Diener, Matthias

AU - Du Bois, André Rauber

AU - Pilla, Maurício Lima

PY - 2022

Y1 - 2022

N2 - Software transactional memory (STM) is an abstraction used for thread synchronization that borrows the concept of transactions from databases. It is often easier to use than locks, proving a high-level abstraction for software developers. In current multicore architectures, data locality is an important aspect of STM performance. Sharing-aware mapping is a technique that aims to improve the performance of applications by mapping threads and data (in the form of memory pages) according to their memory access behavior. In prior work, we successfully used information gained from tracking STM variables and STM operations to perform an effective sharing-aware thread mapping. In this paper, we attempt to extend such a mechanism to perform data mapping. Although initial results using a synthetic application were encouraging, data mapping did not improve performance when using realistic workloads. Contrary to thread mapping, where only keeping track of STM operations is sufficient to perform an effective thread mapping, data mapping requires a global vision of memory page accesses of the application to be able to improve the performance, which STM runtimes can not provide.

AB - Software transactional memory (STM) is an abstraction used for thread synchronization that borrows the concept of transactions from databases. It is often easier to use than locks, proving a high-level abstraction for software developers. In current multicore architectures, data locality is an important aspect of STM performance. Sharing-aware mapping is a technique that aims to improve the performance of applications by mapping threads and data (in the form of memory pages) according to their memory access behavior. In prior work, we successfully used information gained from tracking STM variables and STM operations to perform an effective sharing-aware thread mapping. In this paper, we attempt to extend such a mechanism to perform data mapping. Although initial results using a synthetic application were encouraging, data mapping did not improve performance when using realistic workloads. Contrary to thread mapping, where only keeping track of STM operations is sufficient to perform an effective thread mapping, data mapping requires a global vision of memory page accesses of the application to be able to improve the performance, which STM runtimes can not provide.

KW - Data mapping

KW - NUMA

KW - Sharing-aware

KW - Software transactional memory

UR - http://www.scopus.com/inward/record.url?scp=85129907951&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85129907951&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-04580-6_32

DO - 10.1007/978-3-031-04580-6_32

M3 - Conference contribution

AN - SCOPUS:85129907951

SN - 9783031045790

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

SP - 481

EP - 492

BT - Embedded Computer Systems

A2 - Orailoglu, Alex

A2 - Jung, Matthias

A2 - Reichenbach, Marc

PB - Springer

T2 - 21st International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2021

Y2 - 4 July 2021 through 8 July 2021

ER -