My research interests are mainly related to three subjects:
HLCM is a software components assembly description language that offers a high level of abstraction and thus a potential for good hardware portability while enabling very high performance at execution. HLCM is just an assembly language and it relies on a backend component model for execution, multiple backends have been implemented.
L²C is a software component model designed to be a perfect backend for HLCM. L²C supports components written in C++ and inter-component interactions as local C++ method calls or MPI messages or CORBA RMI.
Gluon++ is a software component model build on top of the Charm++ language and runtime. Gluon++ supports components written in Charm++ or C++ and inter-component interactions as Charm++ enthry method calls or local C++ method calls.
GenericSCA is a prototype demonstrating how to introduce generic programming in a software component model. GenericSCA is based on the Service Component Architecture and extends it with support for generic components.
I work on Gysela5D, a semi-Lagrangian gyrokinetic solver for plasma physics in the field of controlled fusion.
I focus on porting Gysela to the Blue Gene/Q architecture. I also try to modularize the application by applying a software component approach similar to what I proposed during my Ph.D. Thesis.
I worked on using the HLCM component model for various High Performance / High Throughput applications: the MapReduce model, a parallel 2D FFT and a parallel jacobi computation.
I proposed HLCM, a component assembly model in which new composition operators can be introduced. It is especially well suited for High Performance Computing thanks to the minimum overhead it introduces and to the optimizations it enables. It has been validated with experiments on the Grid'5000 platform.
I defended my thesis on December the 6th 2010. The manuscript as well as the slides of my defense are available.
I worked on the support of collective communication à la MPI in software component models.