X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/ThesisAhmed.git/blobdiff_plain/46e8710e38caa11a5ea1ac4d5878d02ebe1cb610..HEAD:/Abstruct.tex?ds=sidebyside diff --git a/Abstruct.tex b/Abstruct.tex index 9aa1432..4c99eb3 100644 --- a/Abstruct.tex +++ b/Abstruct.tex @@ -12,29 +12,33 @@ %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -\emph{ \begin{center} \Large Energy Consumption Optimization of Parallel Iterative Applications using CPU Frequency Scaling\end{center}} +\emph{ \begin{center} \Large Energy Consumption Optimization of Parallel Applications with Iterations using CPU Frequency Scaling\end{center}} %\emph{ \begin{center} \large By \end{center}} -\emph{ \begin{center} \large Ahmed Badri Muslim Fanfakh \\ University of Franche-Comt\'e, 2016 \end{center}} +\emph{ \begin{center} \large Ahmed Badri Muslim FANFAKH \\ University of Franche-Comt\'e, 2016 \end{center}} %\emph{ \begin{center} \large The University of Franche-Comt\'e, 2015 \end{center}} -\emph{ \begin{center} \large Supervisors: Raphaël Couturier, Jean-Claude Charr and Arnaud Giersch \end{center}} +\emph{ \begin{center} \large Supervisors: Raphaël Couturier and Jean-Claude Charr \end{center}} In recent years, green computing has become an important topic in the supercomputing -research domain. However, the computing platforms are still consuming more and more energy due to the increasing number of nodes composing them. +research domain. However, the computing platforms are still consuming more and more energy due to the increase in the number of nodes composing them. To minimize the operating costs of these platforms many techniques have been used. Dynamic voltage and frequency scaling (DVFS) is one of them. It can be used to reduce the power consumption of the CPU while computing, by lowering its frequency. However, lowering the frequency of a CPU may increase -the execution time of an application running on that processor. Therefore, the +the execution time of the application running on that processor. Therefore, the frequency that gives the best trade-off between the energy consumption and the performance of an application must be selected. -In this thesis, a set of works are presented to optimize the energy consumption -and the performance of the synchronous and asynchronous message passing iterative applications running over clusters and grids. The energy consumption and performance models for each type of iterative application -are well modelled and defined according to the characteristics of both the application itself and the parallel architecture executing this application. +This thesis, presents the algorithms developed to optimize the energy consumption +and the performance of synchronous and asynchronous message passing applications with iterations running over clusters or grids. The energy consumption and performance models for each type of parallel application predicts its execution time and energy consumption for any selected frequency +according to the characteristics of both the application and the architecture executing this application. -Firstly, we propose a frequency scaling factors selection algorithm for synchronous iterative applications running over a homogeneous platform. It is applied to the NAS parallel benchmarks and stimulated by SimGrid simulator. -Secondly, we develop two frequency scaling factors selection algorithms for synchronous iterative applications running over a heterogeneous cluster and grid. Both algorithms are implemented to the NAS parallel benchmarks and conducted over SimGrid simulator and Grid'5000 testbed. -Thirdly, we propose a frequency scaling factors selection algorithm for an asynchronous and a hybrid iterative applications running over a grid. The algorithm is evaluated over SimGrid simulator and Grid'5000 testbed while running a multi-splitting application that solves 3D problem. -All the proposed algorithms are compared to an existing methods, which are the Rauber and Rünger and the energy and delay products (EDP) methods. The comparison results show that all the proposed algorithms give better energy consumption and performance trade-off results. The proposed algorithms have a very small overhead on the execution time of the applications and they work online without training and profiling. -\textbf{KEY WORDS:} Dynamic voltage and frequency scaling, Grid computing, Energy optimization, iterative applications and frequency scaling online algorithm. \ No newline at end of file +The contribution of this thesis can be divided into three parts: Firstly, optimizing the trade-off between +the energy consumption and the performance of the message passing applications with synchronous iterations +running over homogeneous clusters. Secondly, adapting the energy and performance models to heterogeneous platforms where each node can have different specifications such as computing power, energy consumption, available frequency gears or network's latency and bandwidth. The frequency scaling algorithm was also modified to suit the heterogeneity of the platform. Thirdly, the models and the frequency scaling algorithm were completely rethought to take into considerations the asynchronism in the communication and computation. +All these models and algorithms were applied to message passing applications with iterations and evaluated +over either SimGrid simulator or Grid'5000 platform. The experiments showed that the proposed algorithms are +efficient and outperform existing methods such as the energy and delay product. They also introduce a small +runtime overhead and work online without any training or profiling. + +\textbf{KEY WORDS:} Dynamic voltage and frequency scaling, Grid computing, Energy optimization, parallel applications with iterations and online frequency scaling algorithm. \ No newline at end of file