Hello Christelle. Could you briefly introduce yourself and describe your background?
I'm originally from Brittany, from Concarneau. I'm fifty years old. After a scientific baccalaureate at the Lycée Brizeux in Quimper, I chose the scientific path that led me to a DEA in materials physics at UBO (Université de Bretagne Occidentale). Quantum physics was an integral part of my training at the time, and I had no idea that years later it would be at the heart of my business. I'm sensitive to environmental issues and attached to my region. I then began a thesis in spatial oceanography at IFREMER. I wasn't able to pursue this project due to a lack of funding. So I had to face up to the situation and choose a different way forward. What seemed like a difficulty turned out to be an opportunity: I wanted to combine science and travel, to discover a world that was constantly changing and that I was unfamiliar with. So I joined CGG (Compagnie Générale de Géophysique), which specializes in seismic data processing. As their software developer, I taught myself Fortran 77, the emblematic language of the scientific community. After 3 years, I wanted to understand the super-machines that use digital simulation to map soil. One of the challenges was to discover raw materials such as oil and gas.
Supercomputing for science has opened a second page in your professional life?
This page goes beyond even my professional life. In 2000, I decided to work in the administration of these large-scale computing systems. The opportunity arose within CGG to go to Angola to manage the IT resources of a data processing center. Angola was in the midst of a civil war at the time, with a lot at stake in the oil and diamond trade. I lived in a house in the city of Luanda, among the Angolan people. I've always refused the tyranny of fear. Immersed in a culture radically different from my own, I was driven to learn how the people around me lived or even survived. I learned Portuguese, their language, to communicate with the center's operators, including a young woman, Luisa, from whom I learned a lot. I also saw the ravages of poverty and experienced the horror of their war. But I also lived through the most instructive years of my life, the ones that open the doors to tolerance, open-mindedness and benevolence.
Professionally, our team made sure that the computing resources were operational H24. Like oil wells, the machine had to keep going. It was my job to see to that. We had to optimize its use so that studies could be delivered to the center's geophysicists on time. This meant optimizing energy consumption. I helped local operators understand these issues, which is not so easy in a country where everything is lacking in everyday life. I then returned to France, where I was lucky enough to join a technology watch team within CGG, always with optimization as the watchword, and particularly on the data storage aspects of an ECU.
In 2008, I left CGG and took time out for my pregnancy and my daughter Sasha's first months. This was essential. After 20 months, I joined SGI before working for the French company Bull, which later became Atos, today one of the leaders in supercomputer design. I flourished there as a supercomputer architect, with the same question for every customer, whether academic or industrial: what will be the most efficient architecture for their applications? This applies to meteorology, aeronautics, the automotive industry and many other fields of research. It was also with Atos that I reactivated my quantum fiber to promote a quantum emulator developed by Atos teams to enable scientific communities to appropriate quantum algorithms while waiting for the first error-tolerant quantum computers.
So the core of your business makes a decisive contribution to science and innovation?
Before the 2000s, we were working with systems whose power didn't exceed a few Teraflops. Then we saw the democratization of computing clusters, based on scalar architectures equipped with powerful interconnection networks that enabled HPC to reach computing powers that no one would have dared imagine 40 years earlier. The only problem was that these machines consumed a lot of electricity, and a way had to be found to enable researchers to continue advancing their work while reducing this consumption. Just over 5 years ago, the idea of using graphics cards initially dedicated to video games emerged for intensive computing use, and is now beginning to be democratized in the world of digital simulation and HPC.
Innovation also means not stopping at what we know, and daring to discover paradigms we don't know. We need to explore them to respond to unsolved problems and open up the field of possibilities. The development of quantum computing is a response to this. It is based on the natural properties of the particles that surround us, photons and electrons for example. We shouldn't be afraid to exploit this new paradigm, even if it's still in its infancy today, and combine it with what we already know - HPC and AI - to make it a joint force. The acronym HPC/Q/ AI certainly has a great future.
From my point of view, while it's not currently possible to predict with certainty what these new paradigms will bring, it's essential to explore them to find out, even if it means making mistakes or not getting out of them what we expected.
As it stands, quantum computing is destined to become a complement, not a substitute, enabling us to solve things we haven't yet been able to solve, notably because of problems of scale, which we can't deal with on conventional machines. Most optimization-intensive problems could be impacted. But don't be fooled by the fact that quantum computing won't be able to solve every problem. Quantum computing does not operate on data, but on particle states.
Finally, I'd like to emphasize here that GENCI's role in deploying this revolution in France and Europe has been and remains decisive, in the HPC-IA/QC approach in particular.
Precisely, how would you describe GENCI's role and your experience here?
GENCI is an essential body for French and European research. It has a leadership role on a par with the very largest research centers in Europe, ensuring not only the provision of traditional computing resources such as HPC, but also innovative computing resources for researchers such as AI and quantum computing. This is crucial to enable our researchers/developers to appropriate these instruments and methods, which will expand the scope of tomorrow's possible discoveries. But GENCI's strength also lies in its people, through its mindset, its relationship with computing centers and all the players in an ecosystem that includes researchers, developers, suppliers, startups, academics, industrialists, public authorities, scientific bodies, students...
In our field, it's necessary to be creative and constantly on the lookout for researchers' needs, even to the point of being an initiator or even a pioneer. In this sense, GENCI was one of the first entities in Europe, if not the first, to give academic and industrial researchers the means to work with the methods and tools of artificial intelligence converged with HPC. It is also doing so by participating in the national HQI program, supporting the implementation of the quantum revolution, in France, and also in Europe.
I've been extremely fortunate for 3 years to participate in the implementation of GENCI's roadmap on all these subjects and to be as close as possible to users to better understand their needs. Today, I've tackled another facet of this value chain: the world of semiconductors, which lies at the heart of computing technologies.
You were involved in a project that recently received the Gordon Bell Award. What is this project? What is its impact? What is the Gordon Bell Prize?
In a few words, the Gordon Bell Prize is an award that recognizes an important contribution to scientific research from the world of computer simulation.
The prize is named after an American researcher in electronics and computer science (Chester Gordon Bell, b. 1934), who worked in particular on various chip microarchitectures. The Gordon Bell prize is awarded each year to a scientific team at the time of the largest gathering of the HPC community, Supercomputing, in the USA. This year, it was an international consortium featuring French (CEA/GENCI), American (Berkeley) and Japanese (Riken) talent that won the award. This project has enabled work on laser-based particle gas pedal modeling to move forward, thanks to the power of the world's largest supercomputers: Frontier, the first machine to boast exaflopic power, Riken's Fugaku, Permulter (NERSC) and Summit. These four machines all have different but complementary technological characteristics. I think it's essential to maintain a high level of technological diversity in the field of digital simulation and AI.
There will be many applications for this joint research, notably in the field of healthcare with flash radiotherapy, which can destroy cancer cells while sparing healthy tissue.In any case, this project is a fine demonstration of the complementary nature of the 2 worlds, science and IT, and of the importance of using digital simulation as a tool in the service of scientific advances, some of which have considerable societal stakes. What's more, it was the alliance of skills across the entire value chain that enabled us to achieve this objective, with researchers, developers, application experts from technology integrators and vendors, and people motivated to move a project forward in a short space of time, since the "bulk" of the work was completed in 4 months, with 3 women taking part in the project, 2 of them French and one Japanese.
The technological and scientific worlds, are renowned for presenting access difficulties for girls and women. What are your experiences, impressions, analysis and advice on this point?
The world of HPC, AI and quantum is made up of passionate people. Many of them have stars in their eyes. I've been very fortunate throughout my career to meet exceptional men and women who have supported me and always given me the desire to move forward. That's what I do every day, working alongside people who put the best of digital technology at the service of science.
To tell the truth, I'm a universalist. I see myself first and foremost as a contributor. My main complex, if I had to have one, would have been my height. I'm not very tall. To achieve my goals, I had to stand out. I've given myself the means to make the most of my opportunities, by working hard and putting my energy into science and computing. I also recognize that this luck, to put it this way, is also rooted in the trust I give and receive. My experience stems from encounters with passionate and caring people. It's an opportunity and a state of mind. And women like others, to quote René Char, must be able to "impose their luck, squeeze their happiness and go towards their risk. "They'll get used to looking at you," he continues. For years, I didn't want people to look at me, but at my skills, and I didn't express my femininity very much. Solidarity, particularly feminine solidarity, and accomplishment helped me to see things differently.
I'm not an advice person. I would say that making science instruction more playful could help make it more attractive. It would be good to illustrate and give meaning to science, and in particular to express in simple words what it can achieve, and to do so very early on in the education of young children, whether girls or boys.
Very concretely, my daughter is at collège Maurice Ravel in Montfort-L'amaury. Since last year, the college has had an aeronautics section. Each teacher, whatever his or her subject, proposes themes based on aeronautics: it's a discovery that goes through physics lessons, but also company testimonials (Airfrance, Thales), the BIA (Brevet d'initiation Aero), baptisms, accompanied flights, meeting mechanics, how an engine works. In the process, I was able to explain to her the purpose of research centers like CERFACS, ONERA and INRIA in ROUEN, and the digital simulation work linked to this world to optimize engine efficiency with a direct impact on the environment, a phenomenon to which our children are very sensitive. She's only 14 years old. She amazes me.
Being a woman in this very masculine environment: I think it can be difficult. Opening up to others and making your own luck seems to me to be a necessary step forward. Being bold is essential! Dare, even if it means making mistakes. It makes you stronger!