CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures

Akshay Srivatsa; Sven Rheindt; Dirk Gabriel; Thomas Wild; Andreas Herkersdorf

doi:10.1007/978-3-030-27562-4_2

CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures

Akshay Srivatsa, Sven Rheindt, Dirk Gabriel, Thomas Wild, Andreas Herkersdorf

Chair of Integrated Systems

Technical University of Munich

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

4 Scopus citations

Abstract

Embedded system applications, with their inherently limited parallelism, rarely exploit all available processing resources in large DSM-based manycore architectures. In addition, global coherence spanning across all tiles does not scale well. Therefore, we have proposed a region-based cache coherence (RBCC) approach that enables coherence among a selectable cluster of tiles in accordance with application requirements. In this paper, we present a novel RBCC-malloc() extension that transparently tailors coherence to actually shared application working sets at runtime. Further, the design and hardware implementation of a flexibly configurable coherency region manager (CRM) supporting RBCC-malloc() are introduced. We synthesized the CRM on an FPGA for a 64-core system and observed a 57% reduction in BRAM-utilization compared to a global coherence directory for regions with up to 16 cores. Experiments reveal an application acceleration of up to 42% compared to a message passing based implementation. We also demonstrate the advantage of RBCC-malloc() compared to standalone RBCC.

Original language	English
Title of host publication	Embedded Computer Systems
Subtitle of host publication	Architectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings
Editors	Dionisios N. Pnevmatikatos, Maxime Pelcat, Matthias Jung
Publisher	Springer Verlag
Pages	18-33
Number of pages	16
ISBN (Print)	9783030275617
DOIs	https://doi.org/10.1007/978-3-030-27562-4_2
State	Published - 2019
Event	19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019 - Samos, Greece Duration: 7 Jul 2019 → 11 Jul 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11733 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019
Country/Territory	Greece
City	Samos
Period	7/07/19 → 11/07/19

Keywords

DSM systems
On-demand coherence
Scalable coherence

Access to Document

10.1007/978-3-030-27562-4_2

Cite this

Srivatsa, A., Rheindt, S., Gabriel, D., Wild, T., & Herkersdorf, A. (2019). CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures. In D. N. Pnevmatikatos, M. Pelcat, & M. Jung (Eds.), Embedded Computer Systems: Architectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings (pp. 18-33). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11733 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-27562-4_2

Srivatsa, Akshay ; Rheindt, Sven ; Gabriel, Dirk et al. / CoD : Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures. Embedded Computer Systems: Architectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings. editor / Dionisios N. Pnevmatikatos ; Maxime Pelcat ; Matthias Jung. Springer Verlag, 2019. pp. 18-33 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{d880b535b0174b518a4839140b93ac0d,

title = "CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures",

abstract = "Embedded system applications, with their inherently limited parallelism, rarely exploit all available processing resources in large DSM-based manycore architectures. In addition, global coherence spanning across all tiles does not scale well. Therefore, we have proposed a region-based cache coherence (RBCC) approach that enables coherence among a selectable cluster of tiles in accordance with application requirements. In this paper, we present a novel RBCC-malloc() extension that transparently tailors coherence to actually shared application working sets at runtime. Further, the design and hardware implementation of a flexibly configurable coherency region manager (CRM) supporting RBCC-malloc() are introduced. We synthesized the CRM on an FPGA for a 64-core system and observed a 57% reduction in BRAM-utilization compared to a global coherence directory for regions with up to 16 cores. Experiments reveal an application acceleration of up to 42% compared to a message passing based implementation. We also demonstrate the advantage of RBCC-malloc() compared to standalone RBCC.",

keywords = "DSM systems, On-demand coherence, Scalable coherence",

author = "Akshay Srivatsa and Sven Rheindt and Dirk Gabriel and Thomas Wild and Andreas Herkersdorf",

note = "Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2019.; 19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019 ; Conference date: 07-07-2019 Through 11-07-2019",

year = "2019",

doi = "10.1007/978-3-030-27562-4_2",

language = "English",

isbn = "9783030275617",

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

publisher = "Springer Verlag",

pages = "18--33",

editor = "Pnevmatikatos, {Dionisios N.} and Maxime Pelcat and Matthias Jung",

booktitle = "Embedded Computer Systems",

}

Srivatsa, A, Rheindt, S, Gabriel, D, Wild, T & Herkersdorf, A 2019, CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures. in DN Pnevmatikatos, M Pelcat & M Jung (eds), Embedded Computer Systems: Architectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11733 LNCS, Springer Verlag, pp. 18-33, 19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019, Samos, Greece, 7/07/19. https://doi.org/10.1007/978-3-030-27562-4_2

CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures. / Srivatsa, Akshay; Rheindt, Sven; Gabriel, Dirk et al.
Embedded Computer Systems: Architectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings. ed. / Dionisios N. Pnevmatikatos; Maxime Pelcat; Matthias Jung. Springer Verlag, 2019. p. 18-33 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11733 LNCS).

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

TY - GEN

T1 - CoD

T2 - 19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019

AU - Srivatsa, Akshay

AU - Rheindt, Sven

AU - Gabriel, Dirk

AU - Wild, Thomas

AU - Herkersdorf, Andreas

N1 - Publisher Copyright: © Springer Nature Switzerland AG 2019.

PY - 2019

Y1 - 2019

N2 - Embedded system applications, with their inherently limited parallelism, rarely exploit all available processing resources in large DSM-based manycore architectures. In addition, global coherence spanning across all tiles does not scale well. Therefore, we have proposed a region-based cache coherence (RBCC) approach that enables coherence among a selectable cluster of tiles in accordance with application requirements. In this paper, we present a novel RBCC-malloc() extension that transparently tailors coherence to actually shared application working sets at runtime. Further, the design and hardware implementation of a flexibly configurable coherency region manager (CRM) supporting RBCC-malloc() are introduced. We synthesized the CRM on an FPGA for a 64-core system and observed a 57% reduction in BRAM-utilization compared to a global coherence directory for regions with up to 16 cores. Experiments reveal an application acceleration of up to 42% compared to a message passing based implementation. We also demonstrate the advantage of RBCC-malloc() compared to standalone RBCC.

AB - Embedded system applications, with their inherently limited parallelism, rarely exploit all available processing resources in large DSM-based manycore architectures. In addition, global coherence spanning across all tiles does not scale well. Therefore, we have proposed a region-based cache coherence (RBCC) approach that enables coherence among a selectable cluster of tiles in accordance with application requirements. In this paper, we present a novel RBCC-malloc() extension that transparently tailors coherence to actually shared application working sets at runtime. Further, the design and hardware implementation of a flexibly configurable coherency region manager (CRM) supporting RBCC-malloc() are introduced. We synthesized the CRM on an FPGA for a 64-core system and observed a 57% reduction in BRAM-utilization compared to a global coherence directory for regions with up to 16 cores. Experiments reveal an application acceleration of up to 42% compared to a message passing based implementation. We also demonstrate the advantage of RBCC-malloc() compared to standalone RBCC.

KW - DSM systems

KW - On-demand coherence

KW - Scalable coherence

UR - http://www.scopus.com/inward/record.url?scp=85071418371&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-27562-4_2

DO - 10.1007/978-3-030-27562-4_2

M3 - Conference contribution

AN - SCOPUS:85071418371

SN - 9783030275617

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

SP - 18

EP - 33

BT - Embedded Computer Systems

A2 - Pnevmatikatos, Dionisios N.

A2 - Pelcat, Maxime

A2 - Jung, Matthias

PB - Springer Verlag

Y2 - 7 July 2019 through 11 July 2019

ER -

Srivatsa A, Rheindt S, Gabriel D, Wild T, Herkersdorf A. CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures. In Pnevmatikatos DN, Pelcat M, Jung M, editors, Embedded Computer Systems: Architectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings. Springer Verlag. 2019. p. 18-33. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-27562-4_2

CoD: Coherence-on-demand – runtime adaptable working set coherence for DSM-based manycore architectures

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this