DVFS-aware performance and energy model of HPC applications


Location: LAAS-CNRS - Team SARA or IRIT - Team SEPIA
Supervisors: Da Costa Georges (georges.da-costa@irit.fr) / Guérout Tom (tguerout@laas.fr)
Open for 2021 Master 1 or Master 2 degrees.

Power consumption of computers is becoming an major concern. To optimise their power consumption it is necessary to have precise information on the behavior of applications. With this information, it is possible to choose the right frequency of a processor. The speed of some applications is not really impacted by changes of this frequency, while for some application it has an important effect.

The goal of this internship is to model the fine grained behavior of applications and to link this behavior with the impact (on performance and energy) of frequency changes. As it is difficult even for programmers to know this exact behavior, we will obtain it through fine grained monitoring (hardware performance counters, RAPL, …)

This internship will:

Acquire behavior of HPC applications at different frequencies (several datasets are already availble) Model the performance and energy in function of the frequency and monitored values If time allows, propose a scheduling algorithm harnessing the model to select the best frequency

The acquisition will be done on Grid5000 (www.grid5000.fr) using Expetator (https://gitlab.irit.fr/sepia-pub/expetator)

The project is in the context of the ENERGUMEN (https://www.irit.fr/energumen/ funded by ANR) project at IRIT (http://www.irit.fr) in Toulouse in the SEPIA team. For more information contact Georges Da Costa (georges.da-costa@irit.fr) and Tom Guerout (Tom Guerout guerout@laas.fr). The main research topics of the SEPIA team are energy and performance optimisation of datacenter (multi-objective scheduling). Discussions are open on the exact topic as long as it stays coherent with the activity of SEPIA team.

Expected ability of the student :



Fast scheduling under energy and QoS constraints in a Fog computing environment


Location: LAAS-CNRS - Team SARA or IRIT - Team SEPIA
Supervisors: Da Costa Georges (georges.da-costa@irit.fr) / Guérout Tom (tguerout@laas.fr)
> Open for 2021 Master 1 or Master 2 degrees.

Internship context:
The explosion of the volume of data exchanged within today's IT systems, due to an increasingly wide use and by an increasingly wide audience (large organizations, companies, general public etc.), has led for several years to question the architectures used until now. Indeed, for the past few years, Fog computing [1], which extends the Cloud computing paradigm to the edge of the network, has been developing steadily, offering more and more possibilities and thus extending the field of Internet of Things applications. The management of these new architectures, involving a large number of heterogeneous devices and applications, potentially large volumes of data to be processed, has the challenge of proposing innovative and high-performance solutions to improve the stability, fluidity, security and efficiency of the services and applications deployed.

Internship topic:
The objective of this internship is to study a task scheduling solution [2], under energy and Quality of Service (QoS) constraints, integrating a learning phase on the intrinsic characteristics of the tasks to be scheduled, aiming at facilitating the decision making of a meta-scheduler. This first step will lead to an overview of different AI methods for data-mining and/or clustering [3]. The problem addressed through the development of the meta-scheduler, in charge of scheduling each task, will be to evaluate the quality of the obtained solutions according to QoS constraints and objectives. The meta-scheduler will initially be able to embed so-called "greedy" algorithms. Upstream or in parallel, a modeling effort of the studied Fog architecture will be required in order to clearly define the contours of the problem. The tasks will be of the "fast" type, implying a relatively short execution time but heterogeneous characteristics and needs (number of resources, time horizon, priority, or energy budget for example) which will reinforce the constraint of the performance/time ratio of the meta-scheduler and will imply an evaluation of the scalability of the proposed solution. The experimentation phase will be done by simulation. A significant effort to master the SimGrid simulator (https://simgrid.org/) will be required.

Details of the key points:
Expected student abilities:
References :

[1] YI, Shanhe, HAO, Zijiang, QIN, Zhengrui, et al. Fog computing: Platform and applications. In : 2015 Third IEEE workshop on hot topics in web systems and technologies (HotWeb). IEEE, 2015. p. 73-78.
[2] MURTAZA, Faizan, AKHUNZADA, Adnan, UL ISLAM, Saif, et al. QoS-aware service provisioning in fog computing. Journal of Network and Computer Applications, 2020, vol. 165, p. 102674.
[3] VERMA, Manish, SRIVASTAVA, Mauly, CHACK, Neha, et al. A comparative study of various clustering algorithms in data mining. International Journal of Engineering Research and Applications (IJERA), 2012, vol. 2, no 3, p. 1379-1384.