Hardware-accelerated analysis of real-time Networks-on-Chip

Yunfeng Ma; Leandro Soares Indrusiak

doi:10.1016/j.micpro.2017.06.011

Hardware-accelerated analysis of real-time Networks-on-Chip

Yunfeng Ma, Leandro Soares Indrusiak

Computer Science

Research output: Contribution to journal › Article › peer-review

Abstract

A real-time Network-on-Chip (NoC) must guarantee that it is able to execute a set of tasks and deliver the communication packets that they generate, all within the respective deadlines even under a worst-case scenario. End-to-End Response Time Analysis (E2ERTA) is a mathematical formulation that can be used to test whether a particular NoC configuration is able to guarantee the timely execution of tasks and delivery packets. The complexity of E2ERTA calculation increases with the increase of the number of tasks and packet flows, and with the core count of the NoC. This paper presents an approach to accelerate E2ERTA calculations through the use of custom hardware and efficient implementation of its mathematical operations. We explore the performance of the proposed approach, and analyse its effectiveness against the state-of-the-art in the field. The results show a significant improvement in testing NoC guarantees, thus potentially enabling the use of E2ERTA as a fast and guaranteed deterministic admission controller for open and dynamic real-time systems. As a case-study, we integrate the proposed approach to a NoC optimisation framework aiming to accelerate the search for NoC configurations that meet all the NoC's hard real-time requirements.

Original language	English
Pages (from-to)	81-91
Number of pages	10
Journal	Microprocessors and Microsystems
Volume	53
Early online date	15 Jun 2017
DOIs	https://doi.org/10.1016/j.micpro.2017.06.011
Publication status	Published - Aug 2017

Bibliographical note

Access to Document

10.1016/j.micpro.2017.06.011

MICPROv3
© 2017 Published by Elsevier B.V. This is an author-produced version of the published paper. Uploaded in accordance with the publisher’s self-archiving policy.
Accepted author manuscript, 1.53 MBLicence: CC BY-NC-ND

Cite this

@article{fede6b2a256a4555ba8d2e4d6c5fe52a,

title = "Hardware-accelerated analysis of real-time Networks-on-Chip",

abstract = "A real-time Network-on-Chip (NoC) must guarantee that it is able to execute a set of tasks and deliver the communication packets that they generate, all within the respective deadlines even under a worst-case scenario. End-to-End Response Time Analysis (E2ERTA) is a mathematical formulation that can be used to test whether a particular NoC configuration is able to guarantee the timely execution of tasks and delivery packets. The complexity of E2ERTA calculation increases with the increase of the number of tasks and packet flows, and with the core count of the NoC. This paper presents an approach to accelerate E2ERTA calculations through the use of custom hardware and efficient implementation of its mathematical operations. We explore the performance of the proposed approach, and analyse its effectiveness against the state-of-the-art in the field. The results show a significant improvement in testing NoC guarantees, thus potentially enabling the use of E2ERTA as a fast and guaranteed deterministic admission controller for open and dynamic real-time systems. As a case-study, we integrate the proposed approach to a NoC optimisation framework aiming to accelerate the search for NoC configurations that meet all the NoC's hard real-time requirements.",

author = "Yunfeng Ma and {Soares Indrusiak}, Leandro",

note = "{\textcopyright} 2017 Published by Elsevier B.V. This is an author-produced version of the published paper. Uploaded in accordance with the publisher{\textquoteright}s self-archiving policy. ",

year = "2017",

month = aug,

doi = "10.1016/j.micpro.2017.06.011",

language = "English",

volume = "53",

pages = "81--91",

journal = "Microprocessors and Microsystems",

issn = "0141-9331",

publisher = "Elsevier",

}

TY - JOUR

T1 - Hardware-accelerated analysis of real-time Networks-on-Chip

AU - Ma, Yunfeng

AU - Soares Indrusiak, Leandro

PY - 2017/8

Y1 - 2017/8

N2 - A real-time Network-on-Chip (NoC) must guarantee that it is able to execute a set of tasks and deliver the communication packets that they generate, all within the respective deadlines even under a worst-case scenario. End-to-End Response Time Analysis (E2ERTA) is a mathematical formulation that can be used to test whether a particular NoC configuration is able to guarantee the timely execution of tasks and delivery packets. The complexity of E2ERTA calculation increases with the increase of the number of tasks and packet flows, and with the core count of the NoC. This paper presents an approach to accelerate E2ERTA calculations through the use of custom hardware and efficient implementation of its mathematical operations. We explore the performance of the proposed approach, and analyse its effectiveness against the state-of-the-art in the field. The results show a significant improvement in testing NoC guarantees, thus potentially enabling the use of E2ERTA as a fast and guaranteed deterministic admission controller for open and dynamic real-time systems. As a case-study, we integrate the proposed approach to a NoC optimisation framework aiming to accelerate the search for NoC configurations that meet all the NoC's hard real-time requirements.

AB - A real-time Network-on-Chip (NoC) must guarantee that it is able to execute a set of tasks and deliver the communication packets that they generate, all within the respective deadlines even under a worst-case scenario. End-to-End Response Time Analysis (E2ERTA) is a mathematical formulation that can be used to test whether a particular NoC configuration is able to guarantee the timely execution of tasks and delivery packets. The complexity of E2ERTA calculation increases with the increase of the number of tasks and packet flows, and with the core count of the NoC. This paper presents an approach to accelerate E2ERTA calculations through the use of custom hardware and efficient implementation of its mathematical operations. We explore the performance of the proposed approach, and analyse its effectiveness against the state-of-the-art in the field. The results show a significant improvement in testing NoC guarantees, thus potentially enabling the use of E2ERTA as a fast and guaranteed deterministic admission controller for open and dynamic real-time systems. As a case-study, we integrate the proposed approach to a NoC optimisation framework aiming to accelerate the search for NoC configurations that meet all the NoC's hard real-time requirements.

U2 - 10.1016/j.micpro.2017.06.011

DO - 10.1016/j.micpro.2017.06.011

M3 - Article

SN - 0141-9331

VL - 53

SP - 81

EP - 91

JO - Microprocessors and Microsystems

JF - Microprocessors and Microsystems

ER -