Projects

MICROCARD

MICROCARD logo Numerical modeling of cardiac electrophysiology at the cellular scale. MICROCARD builds software that can simulate cardiac electrophysiology using whole-heart models with sub-cellular resolution, on exascale supercomputers. From April 2021 to September 2024 MICROCARD was a project funded by EuroHPC call Towards Extreme Scale Technologies and Applications. Since 1 November 2024 MICROCARD is a Centre of Excellence, also funded by EuroHPC, to consolidate and improve the results of the project. It will be funded for 30 months by the call HORIZON-EUROHPC-JU-2023-COE-03.

Contributors: Vincent Loechner, Stéphane Genaud, Cedric Bastoul, Adilla Susungi, Antoine Pierquin.

Link to project site: https://www.microcard.eu/

Related software: https://opencarp.org


OptiTrust

The OptiTrust framework provides programmers with means of optimizing their programs via source-to-source transformations. OptiTrust leverages static analyses based on a subset of Separation Logic to justify that the transformations applied are semantic-preserving. OptiTrust supports source-to-source tranformations that preserve separation logic derivations, allowing to validate the property that the final optimized code satisfies the same specification as the unoptimized input program.
OptiTrust is funded by the French funding agency ANR.

Contributors: Arthur Charguéraud, Thomas Koehler, Nicolas Magaud, Yanni Lefki.

Link to project and software site: https://github.com/charguer/optitrust


NumPEx

Logo NumPEx Our team participates in the Exa-SofT : HPC software and tools program of the NumPEx PEPR (Priority Research Programs and Equipment) initiative. This program aims at consolidating the exascale software ecosystem by providing a coherent, exascale-ready software stack featuring breakthrough research advances enabled by multidisciplinary collaborations between researchers. The main scientific challenges to address are: productivity, performance portability, heterogeneity, scalability and resilience, performance and energy efficiency.

In this context, our team develops APOLLO - Automatic speculative POLyhedral Loop Optimizer, a compiler framework dedicated to automatic, dynamic and speculative parallelization and optimization of programs' loop nests.

Contributors: Philippe Clauss, Bérenger Bramas, Raphaël Colin, Ugo Battiston.

Related software: https://webpages.gitlabpages.inria.fr/apollo


CAMELIA

Sovereign hardware and software components for AI acceleration. CAMELIA develops modular hardware acceleration components and their optimised software stack, including compilers and runtimes. CAMELIA is a French PEPR funded from 2026 to 2032 by the ASIC and Numeric program agencies, through the ANR funding agency. The CAMUS team coordinates the intermediate representation and compiler work-package of CAMELIA. We develop new compilation techniques that facilitate prototyping AI code optimisations at all abstraction levels, from tensor expressions down to hardware ISAs. The goal is to produce highly optimised AI code for new accelerators without having to rewrite hand-optimised libraries or to redesign optimising compilers each time.

Contributors: Thomas Koehler, Cedric Bastoul.