EN FR
EN FR


Section: New Results

Automatic IPC Profile Analysis to Detect Phases in HPC Application

Participants : Mathieu Stoffel, François Broquedis, Frederic Desprez, Abdelhafid Mazouz [Atos/Bull] , Philippe Rols [Atos/Bull] .

Mathieu Stoffel started his PhD in February 2018 on a CIFRE contract with Atos/Bull. The purpose of this work is to enhance the energy consumption of HPC applications on large-scale platforms. The first phase of the thesis project consists in an in-depth study of the evolution of the metrics characterizing the state of the supercomputer during the execution of a highly parallel application. Indeed, the utilization rates of the different components of the HPC system may demonstrate extreme variations during the execution of the aforementioned application. These variations are sometimes subject to repeat themselves on a regular basis during the application execution. We refer to this phenomena as application "phases". In this context, we already generated precise IPC profiles out of many benchmarks and real-life applications and we worked on a methodology to adapt the CPU frequency based on these profiles. This part of the thesis has been published in an IEEE Cluster workshop (HPCMASPA). Currently, we are working on a detection tool for the application phases. It will implement an automated reconfiguration of the parameters of the HPC system during the execution of the application, in relation with the type of phase being executed. By doing so, the tool will aim at optimizing the energy consumption associated with the execution of the application, by adapting the state of the HPC systems all along the aforesaid execution.