The 2024 annual meeting of Exa-Soft

The second Exa-SofT annual assembly, held on November 7th-8th, 2024 at ENSEEIHT in Toulouse, brought together over 50 participants from academia and industry to share progress on scientific computing applications, strengthen collaborations, and introduce the new members of Exa-SofT.

Exa-SofT (HPC software and tools) is one of the five projects of the NumPEx program. It is concerned with research issues around the development of efficient, portable and scalable software for exaflop-scale supercomputers which are, by nature, of a very large size, and heterogeneous. As such, Exa-SofT deals with the development and use of parallel programming models for accelerators such as GPUs, compile and runtime optimizations, the development of low-level libraries for linear and multilinear algebra, and performance and energy analysis and optimization.

The second Exa-SofT annual assembly took place on 7th and 8th November 2024 in the facilities of the ENSEEIHT school of Toulouse. It gathered 50+ participants from all over France. Experts in scientific computing applications from the industrial and academic world also attended the meeting in a joint effort to foster collaborations. During the two days of the event, eight scientific talks were given by members of Exa-SofT to present the latest results and progress achieved in the different work-packages. Two plenary sessions were also organized to discuss about the consolidation and interoperability of the software packages developed in Exa-SofT on the one hand, and about interactions with application experts around the development of mini-apps on the other hand. One highlight of this meeting was the large participation of young Exa-SofT members (PhD students and post-docs), who had the opportunity to introduce themselves – as well their research – to the audience.

Thursday, 07 November 2024

Friday, 8 November 2024

Attendees

  • Ugo Battinston, Inria
  • Julien Bigot, CEA
  • Alfredo Buttari, CNRS
  • Philippe Clauss, Inria
  • Henri Calandra, Total Energies
  • Terry Cojean, Eviden
  • Raphaël Colin, Inria
  • Benoit Combemale, Inria
  • Jean-Marie Couteyen, Airbus
  • Albert d’Aviau de Piolant, Inria
  • Georges Da Costa, Université de Toulouse
  • Alexandre Denis, Inria
  • Nicolas Ducarton, Inria
  • Pierre Esterie, Inria
  • Ewen Brune, Inria
  • Mathieu Faverge, Inria
  • Thierry Gautier, Inria
  • David Goudin, Eviden

  • Karmijn Hoogveld, Université de Toulouse
  • Catherine Guelque, Telecom SudParis
  • Amina Guermouche, Bordeaux INP
  • Abdou Guermouche, Université de Bordeaux
  • Julien Hermann, CNRS
  • Jerry Lacmou Zeutouo, Université de Picardie
  • Marc Pérache, CEA
  • Raymond Namyst, Université de Bordeaux
  • Nicolas Brieuc, Inria
  • Alena Kopanicakova, Toulouse INP
  • Christian Perez, Inria
  • Luc Giraud, Inria
  • Alejandro Estana, Université de Toulouse
  • Lucas Pernollet, CEA
  • Millian Poquet, Université de Toulouse
  • Jules Risse, Telecom SudParis
  • Julien Vanharen, Inria

  • Alec Sadler, Inria
  • Nicolas Renon, Université de Toulouse
  • Philippe Swartvagher, Inria
  • Samuel Thibault, Université de Bordeaux
  • François Trahay, Telecom SudParis
  • Pierre Matalon, École polytechnique
  • Atte Torri, Université Paris-Saclay
  • Bora Ucar, ENS
  • Pierre-André Wacrenier, Université de Bordeaux
  • Damien Gratadour, Université Paris Cité
  • Petr Vacek, IFPEN
  • Matthieu Robeyns, Université Paris-Saclay
  • Nathalie Furmento, CNRS

© PEPR NumPEx


The third co-design and co-development workshop of Exa-DI on "Artificial Intelligence for HPC@Exscale"

The third co-design/co-development workshop of the Exa-DI project (Development and Integration) of the PEPR NumPEx was dedicated to “Artificial Intelligence for HPC@Exscale” targeting the two topics “Image analysis @ exascale” and “Data analysis and robust inference @ exascale”. It took place on October 2 and 3, 2024 at the Espace La Bruyère, Du Côté de la Trinité (DCT) in Paris.

 

This face-to-face workshop brought together, for two days, Exa-DI members, members of the other NumPEx projects (Exa-MA: Methods and Algorithms for Exascale, Exa-SofT: HPC Software and Tools, Exa-DoST: Data-oriented Software and Tools for the Exascale and Exa-AToW: Architectures and Tools for Large-Scale Workflows), Application demonstrators (ADs) from various research and industry sectors and Experts to discuss advancements and future directions for integration of Artificial Intelligence into HPC/HPDA workflows at exascale targeting the two topics, “Large image analysis” and “Data analysis and robust inference”.

 

This workshop is the third co-design/co-development workshops in the series whose main objective is to promote software stack co-development strategies to accelerate exascale development and performance portability of computational science and engineering applications. This workshop is a little different from the previous two in that it has a prospective character targeting the increasing importance of rapidly evolving AI-driven and AI-coupled HPC/HPDA workflows in “Large images analysis @ exascale” and “Data analysis (simulation, experiments, observation) & robust inference @ exascale”. Its main objectives are first to co-develop a shared understanding of the different modes of coupling AI into HPC/HPDA workflows, second to co-identify execution motifs most commonly found  in scientific applications in order to drive the co-development of collaborative specific benchmarks or proxy apps allowing to evaluate/measure end-to-end performance of AI-coupled HPC/HPDA workflows and finally, to co-identify  software components (libraries, frameworks, data communication, workflow tools, abstraction layers, programming and execution environments) to be co-developed and integrated to improve critical components and accelerate them.

Key sessions included

  • Introduction and Context: Setting the stage for the workshop’s two main topics as well as presenting the GT IA, a transverse action in NumPEx.
  • Attendees Self-Introduction: Allowing attendees to introduce themselves and their interests.
  • Various Sessions: These sessions featured talks on the challenges to tackle and bottlenecks to overcome (execution speed, scalability, volume of data…), on the type, the format and the volume of data currently investigated, on the frameworks or programming languages ​currently used (e.g. python, pytorch, JAX, C++, etc..) and on the typical elementary operations performed on data.
  • Discussions and Roundtables: These sessions provided opportunities for attendees to engage in discussions and share insights on the presented topics in order to determine a strategy to tackle the challenges in co-design and co-development process.

Invited speakers

  • Jean-Pierre Vilotte from CNRS, member of Exa-DI, who provided the introductory context for the workshop.
  • Thomas Moreau from Inria, member of Exa-DoST, presenting the GT IA, a transverse action in NumPEx.
  • Tobias Liaudat from CEA, discussing fast and scalable uncertainty quantification for scientific imaging.
  • Damien Gradatour from CNRS, addressing how building new brains for giant astronomical telescopes with Deep Neural Networks?
  • Antoine Petiteau from CEA, discussing data analysis for observing the Universe with Graviational Waves at low frequency.
  • Kevin Sanchis from Safran AI, addressing benchmarking self-supervised learning methods in remote sensing.
  • Hugo Frezat from Université Paris Cité, presenting learning subgrid-scale models for turbulent rotating convection.
  • Benoit Semelin from Sorbonne Université, discussing simulation-based inference with cosmological radiative hydrodynamics simulations for SKA.
  • Bruno Raffin & Thomas Moreau from Inria, presenting Machine Learning based analysis of large simulation outputs in Exa-DoST.
  • Julián Tachella from CNRS, presenting DeepInverse: a PyTorch library for solving inverse problems with deep learning.
  • Erwan Allys from ENS-PSL, exploring generative model and component separation in limited data regime with Scattering Transform.
  • François Lanusse from CNRS, discussing multimodal pre-training for Scientific Data: Towards large data models for Astrophysics. > en ligne
  • Christophe Kervazo from Telecom Paris, addressing interpretable and scalable deep learning methods for imaging inverse problems.
  • Eric Anterrieu from CNRS, exploring deep learning based approach in imaging radiometry by aperture synthesis and its implementation.
  • Philippe Ciuciu from CEA, addressing Computational MRI in the deep learning era.
  • Pascal Tremblin from CEA, characterizing patterns in HPC simulations using AI driven image recognition and categorization.
  • Bruno Raffin from Inria, member of Exa-DI, presenting the Software Packaging in Exa-DI

Outcomes and impacts

Many interesting and fruitful discussions took place during this prospective workshop. These discussions allowed us first to progress in understanding the challenges and bottlenecks underpinning AI-driven HPC/HPDA workflows most commonly found in the ADs. Then, a first series of associated issues to be addressed have been identified and these issues can be gathered in two mains axes: (i) image processing of large volumes, images resulting either from simulations or from experiments and (ii) exploration of high-dimensional and multimodal parameter spaces.

One of the very interesting issues that emerged from these discussions concerns the NumPEx software stack and in particular, how could the NumPEx software stack be increased beyond support for classic AI/ML libraries (e.g. TensorFlow, PyTorch) to support concurrent real time coupled execution of AI and HPC/HPDA workflows in ways that allow the AI systems to steer or inform the HPC/HPDA task and vice versa?

A first challenge is the coexistence and communication between HPC/HPDA and AI tasks in the same workflows. This communication is mainly impaired by the difference in programming models used in HPC (i.e., C++, C; and Fortran) and AI (i.e., Python) which requires a more unified data plane management in which high-level data abstractions could be exposed and to hide from both HPC simulations and AI models the complexities of the format conversion and data storage and data storage and transport. A second challenge concerns using the insight provided by the AI models and simulations for identifying execution motifs commonly found in the ADs to guide, steer, or modify the shape of the workflow by triggering or stopping new HPC/HPDA tasks. This implies that the workflow management systems must be able to ingest and react dynamically to inputs coming from the AI models. This should drive the co-development of new libraries, frameworks or workflow tools supporting AI integration into HPC/HPDA workflows.

In addition, these discussions highlighted that an important upcoming action would be to build cross-functional collaboration between software and workflow components development and integration with the overall NumPEx technologies and streamline developer and user workflows.

 

It was therefore decided during this workshop the set-up of a working group addressing these different issues and allowing in fine the building of a suite of shared and well specified proxy-apps and benchmarks, with well-identified data and comparison metrics addressing these different issues. Several teams of ADs and experts have expressed their interest in participating in this working group that will be formed. A first meeting with all interested participants will be organized shortly.

Attendees

  • Jean-Pierre Vilotte, CNRS and member of Exa-DI
  • Valérie Brenner, CEA and member of Exa-DI
  • Jérôme Bobin, CEA and member of Exa-DI
  • Jérôme Charousset, CEA and member of Exa-DI
  • Mark Asch, Université Picardie and member of Exa-DI
  • Bruno Raffin, Inria and member of Exa-DI and Exa-DoST
  • Rémi Baron, CEA and member of Exa-DI
  • Karim Hasnaoui, CNRS and member of Exa-DI
  • Felix Kpadonou, CEA and member of Exa-DI
  • Thomas Moreau, Inria and member of Exa-DoST
  • Erwan Allys, ENS-PSL and application demonstrator
  • Damien Gradatour, CNRS and application demonstrator
  • Antoine Petiteau, CEA and application demonstrator
  • Hugo Frezat, Université Paris Cité and application demonstrator
  • Alexandre Fournier, Institut de physique du globe and application demonstrator
  • Tobias Liaudat, CEA
  • Jonathan Kem, CEA
  • Kevin Sanchis, Safran AI
  • Benoit Semelin, Sorbonne Université
  • Julian Tachella, CNRS
  • François Lanusse, CNRS
  • Christophe Kervazo, Telecom Paris
  • Eric Anterrieu, CNRS
  • Philippe Ciuiciu, CEA
  • Pascal Tremblin, CEA

© Valérie Brenner


Brice Goglin and Samuel Thibault: HPC mapping experts awarded by the french Académie des sciences

Article originally published on Inria website here

Every November, the french Académie des sciences unveils its prestigious awards. This year, alongside Brice Goglin, the NumPEx member Samuel Thibault have just been awarded the Inria – Académie des sciences – Dassault Systèmes Innovation Prize.

Samuel Thibault is responsible for the runtime part of the Exa-SofT software stack. More specifically, he is in charge of the integration of communications with task scheduling, high-level expression of the division of computational tasks, and fault tolerance. Samuel is also co-leader of the working group Accelerator.

Learn more about his research in the article published by Inria.

Photo credit académie des sciences – Mathieu Baumer


nuage cloud data

Call for proposals "Numérique pour l’Exascale"

The NumPEx program is launching its first call for projects to support advances in high-performance computing (HPC), high-performance data analysis (HPDA) and artificial intelligence (AI). Our France 2030 research program aims to develop software capable of operating future exascale machines, and to prepare the main scientific and industrial application codes.

This call is structured around three axes:

  • Emerging AI methods, algorithms and software for scientific computing and HPC for AI.
  • Programming models adapted to accelerated architectures.
  • Workflows for scientific data analysis, with the SKA project as a use case.

This call for projects has a budget of 4 million euros. It will fund 1 to 2 projects per axis, for a maximum duration of 48 months.

The amount of funding requested must be a minimum of 500 k€ and a maximum of 1 M€, depending on the theme of the project.

The same project manager can only be responsible for one PEPR project, including targeted projects.

Deadline for applications: April 1, 2025.

For more information and to apply

Josselin Garnier appointed to the French Académie des Sciences

Article originally published on the french Académie des sciences website here

Josselin Garnier, professor at École Polytechnique and member of Exa-MA, is one of 18 scientists appointed to the French Académie des sciences in recognition of his significant contributions to research.

Alongside Clément Gauchy, CEA research staff member, Josselin Garnier coordinates the workpackage “Uncertainty quantification” of the Exa-MA project. The objective of this workpackage is to leverage exascale computing to implement advanced analysis methods that require a large number of executions of complex simulation codes to quantify uncertainties in these simulations and solve inverse or optimization problems in the presence of uncertainties. The work carried out aims to quantify the confidence that can be placed in the predictions derived from these simulations and to facilitate decision-making based on them.

Learn more about his career and research in the article published by the French Académie des sciences.

Photo credit Adobe Stock


The 2024 annual meeting of Exa-DoST

The second annual meeting of the NumPEx Exa-DoST project took place at Inria, in Rennes, on 18-19 September 2024. This two days were an opportunity to share the first outcomes of the projet and discuss the upcoming research.

The Exa-DoST project is one of the five targeted projects of the NumPEx program, focusing on the challenges posed by data storage, processing, and analytics in the context of the emergence of Exascale Computing in Europe. The goal is to leverage innovative storage technologies and support complex hybrid workflows involving simulation, analytics, and learning, running at extreme scales across supercomputers interconnected to Clouds and Edge-based systems.

The second annual meeting of the Exa-DoST project took place at Inria, in Rennes, on 18-19 September 2024. It gathered participants in the project (scientists, engineers, students) with the goal of providing a common, shared vision on the project’s objectives, activities, and strategy. It also included specific workshops focused on identifying challenging application requirements in terms of data storage and analytics. For the first time, members of the ExaDoST Scientific Advisory Committee and of the Industrial and Technology Advisory Committee attended a global Exa-DoST meeting.

Wednesday, 18 September 2024

The first day was dedicated to presentations providing an overview of the project’s goals and activities in the various specific areas covered by the technical work packages. These were followed by a discussion on how the activities performed in these work packages (WP) could interact together.

    • Introduction to NumPEx and Exa-DoST
      by Gabriel Antoniu, Inria research scientist and Exa-DoST co-leader
      and Julien Bigot, CEA research scientist and Exa-DoST co-leader
    • Goals and progress status of  the work packages:
      • WP1 – Storage and I/O
        by Francieli Boito, Inria research scientist and Exa-DoST WP leader
        and François Tessier, Inria research scientist and Exa-DoST WP leader
      • WP2 – In situ processing
        By Yushan Wang, CEA research scientist and Exa-DoST WP leader
        and Laurent Colombet, CEA research scientist and Exa-DoST WP leader
      • WP3 – ML-based analytics
        by Thomas Moreau, Inria research scientist and Exa-DoST WP leader
        and Bruno Raffin, Inria research scientist and Exa-DoST WP leader
      •  WP4 – Application illustrator :
        by Virginie Grandgirard, CEA research scientist and Exa-DoST WP leader
        and Damien Gratadour, Université Paris Cité professor and Exa-DoST WP leader

  • Session on inter-WP interaction
    animated by Gabriel Antoniu and Julien Bigot

Thursday, 19 September 2024

The second day was organized as a series of working groups which took place in parallel. The first series of three working groups was dedicated to identifying and characterizing data-related challenges posed by three reference application areas targeted by Exa-DoST: SKA (radio astronomy), Gysela-X (plasma simulation), and Coddex (continuum mechanics).
Brainstorming sessions:
A second series of three working groups was then organized by technical topics (WP1: storage, WP2: in situ processing, WP3: data analysis based on machine learning), with the goal of identifying common application motifs shared by the three applications.
A synthesis of this collective work was presented in a plenary session.
Feedback on application motifs sessions:
    • WP1
      by Francieli Boito and François Tessier
    • WP2
      by Yushan Wang and Laurent Colombet
    • WP3
      by Thomas Moreau and Bruno Raffin

The participants found the exchanges particularly fruitful and decided to continue this work through a series of online and physical meetings.

See the full program

Attendees

  • Gabriel Antoniu, Inria and Exa-DoST leader
  • Rosa Badia, BSC and board member of Exa-DoST
  • Thomas Badts, Inria
  • Andres Bermeo Marinelli, Inria and member of Exa-DoST
  • Julien Bigot, CEA and Exa-DoST leader
  • Francieli Boito, Inria and Exa-DoST WP leader
  • Silvina Caino-Lores, Inria and member of Exa-DoST
  • Damien Chapon, CEA and member of Exa-DoST
  • Laurent Colombet, CEA and Exa-DoST WP leader
  • Almuhisen Feda, CEA and member of Exa-DoST
  • Chiara Ferrari, Observatoire de la Côte d’Azur and Board member of Exa-DoST
  • Virginie Grandgirard, CEA and Exa-DoST WP leader
  • Damien Gratadour, Université Paris Cité and Exa-DoST WP leader
  • Gabriel Hautreux, CINES and board member of Exa-DoST
  • Nicolas Lardjane, CEA and board member of Exa-DoST
  • Pierre-François Lavallée, CNRS and board member of Exa-DoST
  • Jakob Luettgau, Inria and member of Exa-DoST
  • Benoit Martin, CEA and member of Exa-DoST
  • François Mazen, Kitware and board member of Exa-DoST
  • Yann Meurdesoif, CEA and board member of Exa-DoST
  • Dorian Midou, CEA and member of Exa-DoST
  • Shan Mignot, CNRS and member of Exa-DoST
  • Julien Monniot, Inria and member of Exa-DoST
  • Thomas Moreau, Inria and Exa-DoST WP leader
  • Sai Narasimhamurthy, ParTec and board member of Exa-DoST
  • Etienne Ndamlabin, Inria and member of Exa-DoST
  • Jean-Francois Nezan, INSA
  • Thomas Noël, ANR and Exa-DoST point of contact
  • Kevin Obrejan, CEA and member of Exa-DoST
  • Guillaume Pallez, Inria and member of Exa-DoST
  • Lucas Pernollet, CEA and project manager of NumPEx
  • Cédric Prigent, Inria and member of Exa-DoST
  • Abhishek Purandare, Inria and member of Exa-DoST
  • Bruno Raffin, Inria  and Exa-DoST WP leader
  • Stéphane Requena, Genci and board member of Exa-DoST
  • Kento Sato, Riken and board member of Exa-DoST
  • Frederic Suter, CNRS and board member of Exa-DoST
  • François Tessier, Inria and Exa-DoST WP leader
  • Sunrise Wang, Observatoire de la Côte d’Azur
  • Yushan Wang, CEA and Exa-DoST WP leader

© Lucas Pernollet


2023 Inria annual report: NumPEx, a programme aimed at boosting the capacities of exascale computing

The annual report of the French National Institute for Research in Digital Science and Technology (Inria), published last June, provides an overview of the research activities and results of its teams and laboratories. Appointed coordinator of the “Numérique logiciel” program agency in 2024, Inria is a key player in French computer science research. Inria works to ensure that France is part of the European dynamic through the research and innovation of its project teams and collaborations with other research organizations.

Inria’s highlights for 2023 include a panel of France 2030 research programs co-piloted by Inria, including NumPEx!

Find out more on page 10 of the 2023 annual report.

© Guillaume Martel / CEA


NumPEx Highlighted in GENCI's 2023 Annual Report

In the field of research and innovation, GENCI is a key player in the landscape of high-performance computing (HPC) in France. Established in 2007, its mission is to provide the French scientific community with some of the most powerful HPC resources in the world, including the supercomputers Jean Zay, Joliot Curie, and Adastra. These resources enable scientists to perform complex numerical simulations and analyze massive volumes of data, which are crucial for advancements in various fields such as climatology, particle physics, biology, and much more.

Recently, GENCI published its 2023 activity report, and one of the highlights of the year was the launch of NumPEx!

Find NumPEx on page 20 of the 2023 activity report of GENCI.

© Cyril FRESILLON / IDRIS / CNRS Images


2024 InPEx Workshop

Find all the presentation on InPEx website here

The Barcelona Supercomputing Center and NumPEx were thrilled to gather the InPEx community in Sitges, Spain. From June 17th to June 19th, the workshop brought together around 100 HPC experts from Europe, Japan, and the United States.

This event was the perfect opportunity to discuss about the state of the art, projects and program for Exascale and post-Exascale, to present the lastest achievements of the InPEx working groups since the 2023 Workshop, and to work together about the next steps of InPEx.

If you want to know more, all presentations are available on InPEx website.

Photo credit: Corentin Lefevre/Neovia Innovation/Inria


The first workshop of the NumPEx Accelerator working group

On June 12-13th 2024, the Accelerator working group held the workshop “Programmation GPU” to take a first review of the current situation.

This workshop was the perfect occasion to have a comprehensive overview of the various approaches currently available for an effective use of GPUs, including direct programming, libraries, frameworks, and task-based methods.
The workshop enabled participants to leave with a clear understanding of the advantages and disadvantages of each approach and to benefit from insights and experiences with different codes across these approaches.

You will find below all the presentation materials and video recordings of the day’s events, which were held in French.

Introduction and Context

Both presented by Samuel Thibault, professor at Université de Bordeaux

Overview of GPU approaches

Retex session: feedback and experiences

Contributions to the NumPex program and call for proposals

Title image: © George Kedenburg / Unsplash