My research interests concern mainly visualization techniques that are destined to the analysis of complex applications. I focus on the post-mortem analysis of execution traces, which contains information about the events and the resources of the application. The visualisation methods I propose are based on data and visual aggregations, which reduce trace complexity and provide a scalable representation. These methods are efficient to detect anomalies that occur during the execution on multimedia applications on embedded systems, but also in parallel applications performing scientific computing on grids.
Here is a summary of topics I am interested in:
- Performance Visualization Analysis
- Data Aggregation
- Trace Management and Analysis Infrastructures
- Multimedia and Real-Time Embedded Applications
- High-Performance Computing/Distributed Systems/Parallel Applications
The research results I consider most important are available in the publication list below. I encourage you to check and cite these papers in priority.
- A Spatiotemporal Data Aggregation Technique for Performance Analysis of Large-scale Execution Traces. Damien Dosimont, Robin Lamarche-Perrin, Lucas Mello Schorr, Guillaume Huard, Jean-Marc Vincent, IEEE Cluster 2014, pages 149-157, Madrid, September 2014.
- Efficient Analysis Methodology For Huge Application Traces. Damien Dosimont, Generoso Pagano, Guillaume Huard, Vania Marangozova-Martin, Jean-Marc Vincent, 2014 International Conference on High Performance Computing & Simulation, pages 951-958, Bologna, June 2014.
- SoC-Trace 2011-2015 - FUI (France)
- SONGS 2012-2015 - ANR (France)
- LICIA 2011-2015 - CNRS, INRIA, ANR (France) - European Union - CAPES, CNPq, FAPERGS (Brazil)
Publications and Talks
The visualization methods described in my works and publications are implemented in free software tools, described here.