The second co-design and co-development workshop of Exa-DI on "Block-structured AMR @Exascale"
The second co-design/co-development workshop of the Exa-DI project (Development and Integration) of the PEPR NumPEx was dedicated to the computation and communication motif “Block-structured AMR @Exascale”. It took place on February 6 and 7, 2024 at the “Grand Amphi” of the “Institut de Physique du Globe de Paris” 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 block structured AMR at exascale.
This workshop is the second 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. Discussions included challenges in co-design and co-development process, key questions and most urgent issues for collective exploration to build links across NumPEx and the applications, and initiatives promoting exascale software stack sustainability, emphasizing collaboration and innovation.
Key sessions included
- Introduction and Context: Setting the stage for the workshop’s main theme.
- Attendees Self-Introduction: Allowing attendees to introduce themselves and their interests.
- Various Technical Sessions: These sessions featured talks on topics such as exascale performance evaluation and advancements in exascale simulations for different applications like astrophysics simulations, flame fronts and gas/liquid interfaces as well as long molecular dynamic simulations with polarizable force fields. In addition, two experts gave presentations on the Samurai and Hercule libraries and a developer of the massively parallel open-source WarpX Particle-In-Cell code presented his feedback on the implementation of AMReX framework.
- Discussions and RoundTables: These sessions provided opportunities for attendees to engage in discussions and share insights on the presented topics.
Invited speakers
- Jean-Pierre Vilotte from CNRS, member of Exa-DI who provided the introductory context for the workshop.
- Maxime Delorme & Arnaud Durocher from CEA, presenting Dyablo, an AMR code for astrophysics simulations in the exascale era.
- Loic Straffela from Ecole Polytechnique, discussing optimizing I/O performance for AMR Code.
- Igor Chollet from Sorbonne Université, presenting ANKH, a scalable alternative to FFT-based approaches for energy computation on accelerator-based exascale architectures.
- Loic Gouarin from Ecole Polytechnique, presenting SAMURAI: Structured Adaptive mesh and Multi Resolution on Algebra of Intervals
- Luca Fedeli from CEA, discussing implementation of AMReX for WaprX, a Particule-In-Cell code for the exascale era.
- Vincent Moureau from CNRS addressing Dynamic Mesh Adaptation of massive unstructured grids for the simulation of flame fronts and gas/liquid interfaces.
Outcomes and impacts
A very interesting and stimulating outcome that was discussed and decided during this workshop is the set-up of a working group addressing a suite of shared and well specified proxy-apps and mini-apps for this co-design computation and communication motif. Several teams of ADs have expressed their interest in participating in this working group which is being formed and whose first meeting should take place soon.
The discussions allowed us to determine the different goals of this working group. In particular, the criteria of the common mini-apps and proxy-apps that will be built was defined. They have to (i) represent algorithms, data structures and layouts, and other computational and communication characteristics across the different application demonstrators, (ii) leverage and integrate logical suites of software components (libraries, frameworks, tools), (iii) measure interoperability levels, performance gain and/or trade-off between components, performance portability, scalability and software quality and (iv) develop collaborative and shared continuous integration and benchmarking methodologies with standardized performance tools to guide optimizations, together with reference meta-data and specifications models.
The second main goal of this working group, that is also a main goal of the workshop series, is to identify the human resources and expertise in the Computational and Data Team (CDT) that Exa-DI needs to deploy. In the co-design/co-development process, the CDT will ensure the interface between the NumPEx projects and the AD teams to support the co-design and co-development of the mini-apps and proxy-apps suite, together with reference data models for sharing specifications and benchmarking/testing results.
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
- Mark Asch, Université Picardie and member of Exa-DI
- Julien Bigot, Inria and member of Exa-DI
- Karim Hasnaoui, CNRS and member of Exa-DI
- Christophe Prud’homme, Université de Strasbourg and member of Exa-MA
- Hélène Barucq, Inria and member of Exa-MA
- Isabelle Ramière, CEA and member of Exa-MA
- Vincent Faucher, CEA and member of Exa-MA
- Christian Perez, Inria and member of Exa-MA
- Raymon Namyst, Université de Bordeaux and member of Exa-SoFT
- Alfredo Butari, CNRS and member of Exa-SoFT
- Marius Garenaux, Université de Rennes and member of Exa-AToW
- Olivier Martineau, Université de Rennes and member of Exa-AToW
- Vincent Moureau, CNRS and application demonstrator
- Maxime Delorme, CEA and application demonstrator
- Arnaud Durocher, CEA and application demonstrator
- Allan Sacha, CEA and application demonstrator
- Damien Chapon, CEA and application demonstrator
- Grégoire Doeble, CEA and application demonstrator
- Dominique Aubert, Université de Strasbourg and application demonstrator
- Olivier Marchal, Université de Strasbourg and application demonstrator
- Igor Cholet, Université Paris 13 and application demonstrator
- Jean Philippe Piquemal, Sorbonne Université and application demonstrator
- Louis Lagardère, Sorbonne Université and application demonstrator
- Olivier Adjoua, Sorbonne Université and application demonstrator
- Stefano Frambati, Total Energies and application demonstrator
- Luca Fedeli, CEA
- Loic Strafella, École polytechnique
- Loic Gouarin, CNRS
- Marc Massot, École polytechnique
- Pierre Matalon, École polytechnique
- Geoffroy Lesur, CNRS and member of the PEPR Origins
The first co-design and co-development workshop of Exa-DI on "Efficient Discretisation for PDE@Exascale"
The first co-design/co-development workshop of the Exa-DI project (Development and Integration) of the PEPR NumPEx had the topic “Efficient Discretisation for PDE@Exascale” and took place on November 7 and 8, 2023 at the Amphithéâtre J. Talairach (Neurospin) at CEA Saclay in Gif-sur-Yvette.
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 efficient discretisation of physics-based partial differential equations (PDEs) at exascalein discretizing partial differential equations (PDEs) efficiently for exascale applications.
This workshop is the first co-design/co-development workshops in the series whose main objective is to promote co-software stack development strategies to accelerate exascale development and performance portability of computational science and engineering applications. Discussions included challenges in co-design and co-development process, key questions and most urgent issues for collective exploration building links across NumPEx and the applications, and initiatives promoting exascale software stack sustainability, emphasizing collaboration and innovation.
Key sessions included
- Introduction and Context: Setting the stage for the workshop’s main theme.
- Attendees Self-Introduction: Allowing attendees to introduce themselves and their interests.
- Various Technical Sessions: These sessions featured talks on topics like exascale performance evaluation and advancements in exascale simulations for different applications like durable aircraft prototype, CO2 sequestration, turbomachinery, Earth dynamo simulations, dynamic energy simulation for urban buildings, structural and fluid mechanics simulations, geoscience simulations and finally plasma turbulence simualtions. In addition, an expert does a presentation of Kokkos.
- Discussions and RoundTables: These sessions provided opportunities for attendees to engage in discussions and share insights on the presented topics.
Invited speakers
- Jean-Pierre Vilotte from CNRS, member of Exa-DI who provided the introductory context for the workshop.
- Eric Savin from ONERA, discussing exascale performance evaluation for a durable aircraft prototype.
- Henri Calandra from TotalEnergies, discussing exascale multiphysics simulators for CO2 sequestration and monitoring.
- Christian Trott from SNL, presenting on Kokkos.
- Julien Vanharen & Loic Marechal from Inria, addressing exascale simulations for turbomachinery.
- Nathanaël Schaeffer & Hugo Frezat from CNRS, exploring machine learning applications in Earth dynamo simulations.
- Vincent Chabannes & Christophe Prud’homme from Université de Strasbourg, discussing dynamic energy simulation for urban buildings.
- Olivier Jamond from CEA, presenting a new generation HPC PDE solver targeting industrial applications in structural and fluid mechanics, the MANTA project.
- Soleiman Yousef from IFP Energies nouvelles, discussing performance issues in geoscience applications.
- Virginie GrandGirard from CEA, discussing the GYSELA code for plasma turbulence simulations.
Outcomes and impacts
A very interesting and simulating outcome that was discussed and decided during this workshop is the set-up of a working group addressing a suite of shared and well specified proxy-apps and mini-apps for this co-design computation and communication motif. Several teams of ADs have expressed their interest in participating in this working group which is being formed and whose first meeting should take place next January.
The discussions allowed us to determine the different goals of this working group. In particular, the criteria of the common mini-apps and proxy-apps that will be build was defined. They have to (i) represent algorithms, data structures and layouts, and other computational and communication characteristics across the different application demonstrators, (ii) leverage and integrate logical suites of software components (libraries, frameworks, tools), (iii) measure interoperability levels, performance gain and/or trade-off between components, performance portability, scalability and software quality and (iv) develop collaborative and shared continuous integration and benchmarking methodologies with standardized performance tools to guide optimizations, together with reference meta-data and specifications models.
The second main goal of this working group that is also a main goal of the workshop series is to identify the human resources and expertise in the CDT (Computational and Data Team) that Exa-DI needs to deploy. In the co-design/co-development process, the CDT will ensure the interface between the NumPEx projects and the ADs teams to support the co-design and co-development of the mini-apps and proxy-apps suite, together with reference data models for sharing specifications and benchmarking/testing results.
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
- Mark Asch, Université Picardie and member of Exa-DI
- Julien Bigot, Inria and member of Exa-DI
- Karim Hasnaoui, CNRS and member of Exa-DI
- Christophe Prud’homme, Université de Strasbourg and member of Exa-MA
- Hélène Barucq, Inria and member of Exa-MA
- Guillaume Latu, CEA and member of Exa-MA
- Raymond Namyst, Université de Bordeaux and member de Exa-SoFT
- Joshua Bowen, Inria and member of Exa-DoST
- Christian Robert Trott, Sandia National Laboratories
- Virginie Grandgirard, CEA and application demonstrator
- Youssef Soleiman, IFPEN and application demonstrator
- Stéphane de Chaisemartin, IFPEN and application demonstrator
- Ani Anciaux Sedrakian, IFPEN and application demonstrator
- Julien Vanharen, Inria and application demonstrator
- Loic Marechal, Inria and application demonstrator
- Nathanael Saeffer, CNRS and application demonstrator
- Hugo Frezat, CNRS and application demonstrator
- Savin Eric, Onera and application demonstrator
- Denis Gueyffier, Onera and application demonstrator
- Henri Calandra, Total Energies and application demonstrator
- Stefano Frambati, Total Energies and application demonstrator
- Olivier Jamon, CEA and application demonstrator
- Nicolas Lelong, CEA and application demonstrator
- Vincent Chabanne, Université de Strasbourg and application demonstrator
The world's most powerful supercomputer coming soon? Elon Musk's optimistic bet with Dojo
Tesla’s “Dojo” supercomputer, construction of which began this summer, is introduced as the world’s most powerful future supercomputer… by far. Elon Musk has announced that, with this cutting-edge equipment, he will be able to reach 100 exaflops before the end of 2024, thanks to a billion-dollar investment. An ambitious gamble: today, the most advanced computer is 100 times less powerful.
The American supercomputer Frontier passed the symbolic 1 exaflop mark for the first time in history in June 2022.
Invest twice as much to do a hundred times better? That’s Elon Musk’s challenge with “Dojo”. This supercomputer, whose construction announced by Tesla on July 19, will benefit from an investment of $1 billion over three years. The ambition: to create the world’s most powerful computer. And by far. With 100 exaflops expected for October 2024, the computer will be a 100 times more powerful than the world’s most powerful computer to date. It is destined to drive artificial intelligence models behind self-driving cars.
Surpassed for the first time in 2022 by the American computer Frontier, the exaflop barrier corresponds to the execution of 1 billion of billion floating-point operations per second. Germany and France are set to join the coveted club of exaflop supercomputers, in 2024 and 2025 respectively, with the European supercomputers Jupiter and Jules Verne. Production of the American computer began this summer, with the aim of entering the world’s top 5 at over 0.2 exaflops by January 2024. But the technical reality promises to be more complex.
A tight deadline
“Reaching 100 exaflops by the end of 2024 seems ambitious”, says Jean-Yves Berthou, Director of the Centre Inria in Saclay, France. It took Frontier almost a year to pass the exaflop mark. “Here, we would go from a situation with no machines to 100 exaflops in one year. It’s not impossible, but it’s very optimistic”, agrees François Bodin, professor at the Université de Rennes. The scientist highlights, for example, to the uncertainties surrounding the supply of GPU graphics chips, pantagrual data storage or the supply of electricity. Indeed, the average consumption of a supercomputer would be 20 megawatts, according to the CEA.
Tesla’s announcement also keeps the nature of these 100 exaflops unclear. The company has not specified whether this is a theoretical or actual capacity. The difference can be significant: corrected for factors such as memory latency or inter-node communication, Frontier achieves an actual capacity of 1.1 exaflops for a theoretical peak of 1.7 exaflops. To evaluate computer performance under real-life conditions, the supercomputers run the “Linpack” a test bench that brings together several programs and software libraries. “It’s not the same thing to 100 exaflops by running the Linpack algorithm producing 100 exaflops by running theLinpack algorithm than having a machine with a theoretical peak power producing 100 exaflops by running the Linpack algorithm than having a machine with a theoretical peak power of 100 exaflops”, sums up François Bodin.
A supercomputer like no other
Nevertheless, the major strength of Dojo’s computing power lies in its specialization. Frontier, Jupiter or Jules Verne have a wide range of applications – research in the fields of space, climate, pharmaceuticals, energy and many more. Dojo, on the other hand, has just one: training artificial intelligence models for autonomous cars. This type of calculation requires less precision than scientific calculations.
This difference enables Tesla to build a specialized processor architecture, with a lower number of bits (and therefore a lower precision of calculation). precision). “In all likelihood, they will run on 8-bit computation, rather than 64-bit as in classical scientific computing”, explains Jean-Yves Berthou. They will therefore gain a factor of 8 in power.” In other words, by agreeing to divide its calculation precision by 8, Dojo will be able to carry out 8 times as many instructions. We shouldn’t think that Europe is lagging behind, because it would invest half as much for a hundred times less performance”, argues the scientist. The exaflop metric is simply not used in the same context here.
Tesla’s gamble has already had a significant impact: the rise in its stock market price. A Morgan Stanley report estimates that Tesla’s market capitalization could rise by almost $600 billion (562 billion euros) thanks to the potential of its supercomputer, which is expected to help create robot cabs. As is often the case, Elon Musk’s announcements seduce investors. But beware : the billionaire already said he was “very confident” about the appearance of robot cabs in… 2020. His ambitions for Dojo could be just as over-optimistic.