An open source tool chain for performance analysis

Kevin Coulomb (1) , Augustin Degomme (2) , Mathieu Faverge (3) , François Trahay (4)

1 SysFera [Villeurbanne]
2 LIG - Laboratoire d'Informatique de Grenoble
3 ICL - Innovative Computing Laboratory, Dpt of Electrical Engineering & Computer Science
4 TSP - INF - Département Informatique

Modern supercomputers with multi-core nodes enhanced by accelerators, as well as hybrid programming models, introduce more complexity in modern applications. Efficiently Exploiting all of the available resources requires a complex performance analysis of applications in order to detect time-consuming or idle sections. This paper presents an open-source tool-chain for analyzing the performance of parallel applications. It is composed of a trace generation framework called eztrace, a generic interface for writing traces in multipe format called gtg, and a trace visualizer called vite. These tools cover the main steps of performance analysis - from the instrumentation of applications to the trace analysis - and are designed to maximize the compatibility with other performance analysis tools. Thus, these tools support multiple file formats and are not bound to a particular programming model. The evaluation of these tools show that they provide similar performance compared to other analysis tools, while being generic.

Mots clés: Performance analysis; Parallel application; MPI

Domaines: Calcul parallèle, distribué et partagé [cs.DC]

Kevin Coulomb, Augustin Degomme, Mathieu Faverge, François Trahay. An open source tool chain for performance analysis. 5th Parallel Tools Workshop, Sep 2011, Dresden, Germany. ⟨hal-00707236⟩