GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks

Maxime Mirka; Maxime France France Pillois; Gilles Sassatelli; Abdoulaye Gamatié

doi:10.1145/3444950.3447283

GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks

Maxime Mirka, Maxime France France Pillois, Gilles Sassatelli, Abdoulaye Gamatié

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

Abstract

We propose GANNoC, a framework for automatic generation of customized Network-on-Chip (NoC) topologies, which exploits generative adversarial networks (GANs) learning capabilities. We define the problem of NoC generation as a graph generation problem, and train a GAN to produce such graphs. We further present a Reward-WGAN (RWGAN) architecture, based on the Wasserstein GAN (WGAN). It is coupled to a reward network enabling to steer the resulting generative system towards topologies having desired properties. We illustrate this capability through a case study aimed at producing topologies with a specific number of physical connections. After training, the generative network produces unique topologies with a 36% improvement regarding the number of connections, when compared to those found in the training dataset. NoCs’ performance assessment is carried out using the Ratatoskr 3D-NoC simulator with state-of-the-art characteristics. Results suggest interesting opportunities in learning correlations between intrinsic NoC features and resulting performance.

Original language	English
Title of host publication	Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings
Place of Publication	New York, NY, USA
Publisher	Association for Computing Machinery, Inc
Pages	51–58
Number of pages	8
ISBN (Electronic)	9781450389525
DOIs	https://doi.org/10.1145/3444950.3447283
Publication status	Published - 24 Feb 2021
Event	13th Workshop Rapid Simulation and Performance Evaluation: Methods and Tools 2021 - Budapest, Hungary Duration: 20 Jan 2021 → 20 Jan 2021 Conference number: 13 https://rapidoworkshop.github.io/2021/index.html

Publication series

Name	DroneSE and RAPIDO '21
Publisher	Association for Computing Machinery

Workshop

Workshop	13th Workshop Rapid Simulation and Performance Evaluation: Methods and Tools 2021
Abbreviated title	RAPIDO 2021
Country/Territory	Hungary
City	Budapest
Period	20/01/21 → 20/01/21
Internet address	https://rapidoworkshop.github.io/2021/index.html

Keywords / Materials (for Non-textual outputs)

Network-on-Chip
Generative Adversarial Network
Neural Networks
NoC Topology

Access to Document

10.1145/3444950.3447283Licence: All Rights Reserved

https://dl.acm.org/doi/10.1145/3444950.3447283Licence: All Rights Reserved

Cite this

Mirka, M., France Pillois, M. F., Sassatelli, G., & Gamatié, A. (2021). GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks. In Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings (pp. 51–58). (DroneSE and RAPIDO '21). Association for Computing Machinery, Inc. https://doi.org/10.1145/3444950.3447283

Mirka, Maxime ; France Pillois, Maxime France ; Sassatelli, Gilles et al. / GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks. Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings. New York, NY, USA : Association for Computing Machinery, Inc, 2021. pp. 51–58 (DroneSE and RAPIDO '21).

@inproceedings{2052722f91e448aa9959c7cb1f6af878,

title = "GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks",

abstract = "We propose GANNoC, a framework for automatic generation of customized Network-on-Chip (NoC) topologies, which exploits generative adversarial networks (GANs) learning capabilities. We define the problem of NoC generation as a graph generation problem, and train a GAN to produce such graphs. We further present a Reward-WGAN (RWGAN) architecture, based on the Wasserstein GAN (WGAN). It is coupled to a reward network enabling to steer the resulting generative system towards topologies having desired properties. We illustrate this capability through a case study aimed at producing topologies with a specific number of physical connections. After training, the generative network produces unique topologies with a 36% improvement regarding the number of connections, when compared to those found in the training dataset. NoCs{\textquoteright} performance assessment is carried out using the Ratatoskr 3D-NoC simulator with state-of-the-art characteristics. Results suggest interesting opportunities in learning correlations between intrinsic NoC features and resulting performance.",

keywords = "Network-on-Chip, Generative Adversarial Network, Neural Networks, NoC Topology",

author = "Maxime Mirka and {France Pillois}, {Maxime France} and Gilles Sassatelli and Abdoulaye Gamati{\'e}",

year = "2021",

month = feb,

day = "24",

doi = "10.1145/3444950.3447283",

language = "English",

series = "DroneSE and RAPIDO '21",

publisher = "Association for Computing Machinery, Inc",

pages = "51–58",

booktitle = "Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings",

note = "13th Workshop Rapid Simulation and Performance Evaluation: Methods and Tools 2021, RAPIDO 2021 ; Conference date: 20-01-2021 Through 20-01-2021",

url = "https://rapidoworkshop.github.io/2021/index.html",

}

Mirka, M, France Pillois, MF, Sassatelli, G & Gamatié, A 2021, GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks. in Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings. DroneSE and RAPIDO '21, Association for Computing Machinery, Inc, New York, NY, USA, pp. 51–58, 13th Workshop Rapid Simulation and Performance Evaluation: Methods and Tools 2021, Budapest, Hungary, 20/01/21. https://doi.org/10.1145/3444950.3447283

GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks. / Mirka, Maxime; France Pillois, Maxime France; Sassatelli, Gilles et al.
Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings. New York, NY, USA: Association for Computing Machinery, Inc, 2021. p. 51–58 (DroneSE and RAPIDO '21).

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

TY - GEN

T1 - GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks

AU - Mirka, Maxime

AU - France Pillois, Maxime France

AU - Sassatelli, Gilles

AU - Gamatié, Abdoulaye

N1 - Conference code: 13

PY - 2021/2/24

Y1 - 2021/2/24

N2 - We propose GANNoC, a framework for automatic generation of customized Network-on-Chip (NoC) topologies, which exploits generative adversarial networks (GANs) learning capabilities. We define the problem of NoC generation as a graph generation problem, and train a GAN to produce such graphs. We further present a Reward-WGAN (RWGAN) architecture, based on the Wasserstein GAN (WGAN). It is coupled to a reward network enabling to steer the resulting generative system towards topologies having desired properties. We illustrate this capability through a case study aimed at producing topologies with a specific number of physical connections. After training, the generative network produces unique topologies with a 36% improvement regarding the number of connections, when compared to those found in the training dataset. NoCs’ performance assessment is carried out using the Ratatoskr 3D-NoC simulator with state-of-the-art characteristics. Results suggest interesting opportunities in learning correlations between intrinsic NoC features and resulting performance.

AB - We propose GANNoC, a framework for automatic generation of customized Network-on-Chip (NoC) topologies, which exploits generative adversarial networks (GANs) learning capabilities. We define the problem of NoC generation as a graph generation problem, and train a GAN to produce such graphs. We further present a Reward-WGAN (RWGAN) architecture, based on the Wasserstein GAN (WGAN). It is coupled to a reward network enabling to steer the resulting generative system towards topologies having desired properties. We illustrate this capability through a case study aimed at producing topologies with a specific number of physical connections. After training, the generative network produces unique topologies with a 36% improvement regarding the number of connections, when compared to those found in the training dataset. NoCs’ performance assessment is carried out using the Ratatoskr 3D-NoC simulator with state-of-the-art characteristics. Results suggest interesting opportunities in learning correlations between intrinsic NoC features and resulting performance.

KW - Network-on-Chip

KW - Generative Adversarial Network

KW - Neural Networks

KW - NoC Topology

U2 - 10.1145/3444950.3447283

DO - 10.1145/3444950.3447283

M3 - Conference contribution

T3 - DroneSE and RAPIDO '21

SP - 51

EP - 58

BT - Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings

PB - Association for Computing Machinery, Inc

CY - New York, NY, USA

T2 - 13th Workshop Rapid Simulation and Performance Evaluation: Methods and Tools 2021

Y2 - 20 January 2021 through 20 January 2021

ER -

Mirka M, France Pillois MF, Sassatelli G, Gamatié A. GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks. In Proceedings of the 2021 Drone Systems Engineering and Rapid Simulation and Performance Evaluation: Methods and Tools Proceedings. New York, NY, USA: Association for Computing Machinery, Inc. 2021. p. 51–58. (DroneSE and RAPIDO '21). doi: 10.1145/3444950.3447283

GANNoC: A Framework for Automatic Generation of NoC Topologies Using Generative Adversarial Networks

Abstract

Publication series

Workshop

Keywords / Materials (for Non-textual outputs)

Access to Document

Fingerprint

Cite this