The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud

Christoforos Kachris, Dimitrios Soudris, Stelios Mavridis, Manolis Pavlidakis, Christi Symeonidou, Christos Kozanitis, Angelos Bilas, Damon Fenacci, Sharatchandra Varma Bogaraju, Hans Vandierendonck, Dimitrios S. Nikolopoulos

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Emerging cloud applications like machine learning, AI and big data analytics required high performance computing systems that can sustain the increased amount of data processing without consuming excessive power. Towards this end, many cloud operators have started deploying hardware accelerators, like FPGAs, to increase the performance of computational intensive tasks but increasing the programming complexity to utilize these accelerators. VINEYARD has developed an efficient framework that allows the seamless deployment and utilization of hardware accelerators in the cloud without increasing the programming complexity and offering the flexibility of software packages. This paper presents the main components that have been developed in this framework such as the runtime system, the virtualization and the central acceleratorsâĂŹ repository. The proposed platform has been demonstrated into 2 real use cases; neurocomputing applications and machine learning algorithms. The performance evaluation shows that the proposed scheme can achieve up to 25x speedup without increasing the programming complexity of these applications and it also demonstrates how it can be used for large suite of applications.
LanguageEnglish
Title of host publicationProceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation
Place of PublicationNew York, NY, USA
Pages236-243
Number of pages8
DOIs
Publication statusPublished - 15 Jul 2018

Publication series

NameSAMOS '18
PublisherACM

Fingerprint

Particle accelerators
Hardware
Learning systems
Computer programming
Software packages
Learning algorithms
Computer hardware
Field programmable gate arrays (FPGA)

Keywords

  • FPGAs
  • cloud computing
  • hardware accelerators
  • reconfigurable computing
  • machine learning,

Cite this

Kachris, C., Soudris, D., Mavridis, S., Pavlidakis, M., Symeonidou, C., Kozanitis, C., ... Nikolopoulos, D. S. (2018). The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud. In Proceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation (pp. 236-243). (SAMOS '18). New York, NY, USA. https://doi.org/10.1145/3229631.3236093
Kachris, Christoforos ; Soudris, Dimitrios ; Mavridis, Stelios ; Pavlidakis, Manolis ; Symeonidou, Christi ; Kozanitis, Christos ; Bilas, Angelos ; Fenacci, Damon ; Bogaraju, Sharatchandra Varma ; Vandierendonck, Hans ; Nikolopoulos, Dimitrios S. / The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud. Proceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation. New York, NY, USA, 2018. pp. 236-243 (SAMOS '18).
@inproceedings{491a2da959784d05ac20a73dcb49269b,
title = "The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud",
abstract = "Emerging cloud applications like machine learning, AI and big data analytics required high performance computing systems that can sustain the increased amount of data processing without consuming excessive power. Towards this end, many cloud operators have started deploying hardware accelerators, like FPGAs, to increase the performance of computational intensive tasks but increasing the programming complexity to utilize these accelerators. VINEYARD has developed an efficient framework that allows the seamless deployment and utilization of hardware accelerators in the cloud without increasing the programming complexity and offering the flexibility of software packages. This paper presents the main components that have been developed in this framework such as the runtime system, the virtualization and the central accelerators{\^a}ĂŹ repository. The proposed platform has been demonstrated into 2 real use cases; neurocomputing applications and machine learning algorithms. The performance evaluation shows that the proposed scheme can achieve up to 25x speedup without increasing the programming complexity of these applications and it also demonstrates how it can be used for large suite of applications.",
keywords = "FPGAs, cloud computing, hardware accelerators, reconfigurable computing, machine learning,",
author = "Christoforos Kachris and Dimitrios Soudris and Stelios Mavridis and Manolis Pavlidakis and Christi Symeonidou and Christos Kozanitis and Angelos Bilas and Damon Fenacci and Bogaraju, {Sharatchandra Varma} and Hans Vandierendonck and Nikolopoulos, {Dimitrios S.}",
year = "2018",
month = "7",
day = "15",
doi = "10.1145/3229631.3236093",
language = "English",
isbn = "978-1-4503-6494-2",
series = "SAMOS '18",
publisher = "ACM",
pages = "236--243",
booktitle = "Proceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation",

}

Kachris, C, Soudris, D, Mavridis, S, Pavlidakis, M, Symeonidou, C, Kozanitis, C, Bilas, A, Fenacci, D, Bogaraju, SV, Vandierendonck, H & Nikolopoulos, DS 2018, The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud. in Proceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation. SAMOS '18, New York, NY, USA, pp. 236-243. https://doi.org/10.1145/3229631.3236093

The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud. / Kachris, Christoforos; Soudris, Dimitrios; Mavridis, Stelios; Pavlidakis, Manolis; Symeonidou, Christi; Kozanitis, Christos; Bilas, Angelos; Fenacci, Damon; Bogaraju, Sharatchandra Varma; Vandierendonck, Hans; Nikolopoulos, Dimitrios S.

Proceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation. New York, NY, USA, 2018. p. 236-243 (SAMOS '18).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud

AU - Kachris, Christoforos

AU - Soudris, Dimitrios

AU - Mavridis, Stelios

AU - Pavlidakis, Manolis

AU - Symeonidou, Christi

AU - Kozanitis, Christos

AU - Bilas, Angelos

AU - Fenacci, Damon

AU - Bogaraju, Sharatchandra Varma

AU - Vandierendonck, Hans

AU - Nikolopoulos, Dimitrios S.

PY - 2018/7/15

Y1 - 2018/7/15

N2 - Emerging cloud applications like machine learning, AI and big data analytics required high performance computing systems that can sustain the increased amount of data processing without consuming excessive power. Towards this end, many cloud operators have started deploying hardware accelerators, like FPGAs, to increase the performance of computational intensive tasks but increasing the programming complexity to utilize these accelerators. VINEYARD has developed an efficient framework that allows the seamless deployment and utilization of hardware accelerators in the cloud without increasing the programming complexity and offering the flexibility of software packages. This paper presents the main components that have been developed in this framework such as the runtime system, the virtualization and the central acceleratorsâĂŹ repository. The proposed platform has been demonstrated into 2 real use cases; neurocomputing applications and machine learning algorithms. The performance evaluation shows that the proposed scheme can achieve up to 25x speedup without increasing the programming complexity of these applications and it also demonstrates how it can be used for large suite of applications.

AB - Emerging cloud applications like machine learning, AI and big data analytics required high performance computing systems that can sustain the increased amount of data processing without consuming excessive power. Towards this end, many cloud operators have started deploying hardware accelerators, like FPGAs, to increase the performance of computational intensive tasks but increasing the programming complexity to utilize these accelerators. VINEYARD has developed an efficient framework that allows the seamless deployment and utilization of hardware accelerators in the cloud without increasing the programming complexity and offering the flexibility of software packages. This paper presents the main components that have been developed in this framework such as the runtime system, the virtualization and the central acceleratorsâĂŹ repository. The proposed platform has been demonstrated into 2 real use cases; neurocomputing applications and machine learning algorithms. The performance evaluation shows that the proposed scheme can achieve up to 25x speedup without increasing the programming complexity of these applications and it also demonstrates how it can be used for large suite of applications.

KW - FPGAs

KW - cloud computing

KW - hardware accelerators

KW - reconfigurable computing

KW - machine learning,

U2 - 10.1145/3229631.3236093

DO - 10.1145/3229631.3236093

M3 - Conference contribution

SN - 978-1-4503-6494-2

T3 - SAMOS '18

SP - 236

EP - 243

BT - Proceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation

CY - New York, NY, USA

ER -

Kachris C, Soudris D, Mavridis S, Pavlidakis M, Symeonidou C, Kozanitis C et al. The VINEYARD Integrated Framework for Hardware Accelerators in the Cloud. In Proceedings of the 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation. New York, NY, USA. 2018. p. 236-243. (SAMOS '18). https://doi.org/10.1145/3229631.3236093