Pallas: a generic trace format for large HPC trace analysis. Catherine Guelque, Valentin Honoré, Philippe Swartvagher, Gaël Thomas, François Trahay. In IPDPS 2025.[bib][pdf]
Fine-grain energy consumption modeling of HPC task-based programs. Jules Risse, Amina Guermouche, François Trahay. Cluster 2025.[bib][pdf]
Predicting GPU kernel's performance on upcoming architectures. Lucas Van Lanker, Hugo Taboada, Elisabeth Brunet, François Trahay. In Euro-Par, 2024.[bib][pdf]
PYTHIA: an oracle to guide runtime system decisions. Alexis Colin, François Trahay, Denis Conan. In IEEE Cluster 2022.[bib][pdf]
Using differential execution analysis to identify thread interference. Mohamed Saïd Mosli Bouksi, François Trahay, et al. In IEEE TPDS.[bib][pdf]
NumaMMA: Numa MeMory Analyzer. François Trahay, Manuel Selva, Lionel Morel, Kevin Marquet. In ICPP 2018.[bib][pdf]
EZTrace: a generic framework for performance analysis. François Trahay, François Rue, Mathieu Faverge, Yutaka Ishikawa, Raymond Namyst and Jack Dongarra. In CCGrid 2011.[bib][pdf]
As part of my research, I developped multiple pieces of software that are available as open source:
EZTrace is tracing tool for HPC (High Performance Computing) programs. It runs parallel application, intercepts function calls, and generates execution trace files that can be interpreted by visualization tools such as ViTE. If you want to understand your application performance, just run it with EZTrace and use the ViTE (Visual Trace Explorer) open source tool to visualize the program behavior.
Example of execution trace of a MPI+OpenMP program (generated with EZTrace)
Pallas is a library for storing events generated by tracing tools like EZTrace. It is designed for speeding up the post-mortem analysis of traces.
NumaMMA is a memory profiler. The profiler allows to run an application while gathering information about memory accesses. The analyzer visually reports information about the memory behavior of the application allowing to identify memory access patterns.
Memory accesses over time, by each thread, to the pages of cvar_ in NPB LU
I have currently involved in the following research projects:
François Trahay is a full professor at Télécom SudParis since 2022. He is a member of the Benagil Inria team of the computer science department, which investigates high-performance systems. He received his Ph.D. degree in computer science from the University of Bordeaux in 2009, and was an associate professor at Télécom SudParis from 2011 to 2022. He has been working on runtime systems for high performance computing since 2006. His research interests now mostly focus on performance analysis for HPC and distributed systems. He is the project leader and main developper of the EZTrace framework for performance analysis. He is the co-author of research papers in top international conferences, workshops and journals such as IEEE TPDS, IPDPS, ICPP, IEEE Cluster, ACM TACO, or IJPP.
