Section: Partnerships and Cooperations

Regional Initiatives

Idex Prim'Eau

Participant : Jens Gustedt [contact] .

In the framework of the Prim'Eau project of the University of Strasbourg, we study surface runoff for hydrological periods of several days. We use an efficient domain decomposition method that we apply to a real world example of Mutterbach (Moselle) with geological and flood data from the years 1920, 1940 and 2017. As the time and memory usage for these computations is important, we aim to parallelize them.

ADT ASNAP

Participants : Philippe Clauss, Jens Gustedt, Maxime Mogé.

Philippe Clauss, Jens Gustedt and Maxime Mogé have been involved until August 2018 in the ADT Inria project ASNAP (Accélération des Simulations Numériques pour l’Assistance Peropératoire), in collaboration with the Inria team MIMESIS. The goal was to find opportunities in the SOFA simulation platform for applying automatic parallelization techniques developed by Camus. We have investigated two approaches. One approach uses memory behavior memoïzation to generate a parallel code made of independent threads at runtime.

ADT ALTO (ApoLlo TakeOff)

Participants : Philippe Clauss, Muthena Abdul Wahab.

The Apollo compilation platform [4] that is being developed in Camus, dedicated to speculative and dynamic optimization and parallelization of loop nests, is the achievement of many original advances in compilation algorithms, in extensions of the polyhedral model, in speculative parallelization and in dynamic optimization of programs. It is a library of implemented knowledge and a fertile ground for other advances and extensions : for instance, an extension of the polyhedral model for handling non-linear loops would not have been possible without Apollo. However, this software platform must continuously be maintained, improved and extended.

The ALTO project, which is a 2.5 years project started in August 2018, is devoted to strengthen Apollo's software implementation in several ways, thanks to the expert engineer who has been recruited for these goals, Matthew Wahab. The main goals are the following:

• making the programming code respecting the standard rules of open-source software;

• making Apollo more robust regarding cases where some inputs may yield extreme behaviors

• implementing required improvements and extensions, as inter-procedural analysis or memory behavior memoïzation.