Video: Bull Atos Announces Exascale Program at SC14

Print Friendly, PDF & Email

In this video from SC14, Pascale Bernier-Bruna from Bull/Atos describes the company’s ambitious plan to deliver Exascale levels of high performance computing in the next decade.

The main ambition of this Exascale program is to design and develop in Europe the next generation of supercomputers that will contribute to producing world-class solutions for both research and industrial purposes. With one of the largest teams dedicated to High-Performance Computing – now able to leverage the international strength of Atos, with its 86,000 employees worldwide – Bull is experiencing a paradigm shift that will give it the critical mass to address vertical market sectors across the entire world. Exascale will increase the competitiveness and innovation capabilities of industries and address the industrial, scientific and societal challenges of the 2020s, from nanoscience and genomics to climatology, aeronautics and energy.”

Full Transcript:

insideHPC: Good to see you again. You guys made an announcement this week that involved exascale. What is that all about?

Pascale Bernier-Bruna: Yes. Bull made an important– a very ambitious exascale announcement this week about our exascale program. Our objective is to deliver the first exascale systems designed in Europe – designed and developed in Europe. Now there is no miracle technology that will get anybody to the exascale [laughter]. It’s more a matter of working on all of the components of the global super computing environment, and it’s the addition of all these advances that will get us to the exascale. Also, we must keep in mind that exascale is not just about exaFLOPS, it’s also about exabytes of data. I think we’re really getting to the crossroads where HPC needs big data.

insideHPC: These large machines – are we talking 2017, 2025? What is your goal there?

Pascale Bernier-Bruna: I don’t know either [chuckles]. It will also depend on when our customers will get the money. It depends also on partner technology. I don’t think it’s really important. Is it 2020, 2021? You have to get ready now, and that’s what we’re doing.

insideHPC: So there’s a lot of preparation involved, I get that. Pascale, when I think about the global exascale preparations that are going on, I see Europe as a leader on the software side. The Bull Atos company is a hardware vendor last I checked. Help me out.

Pascale Bernier-Bruna: Yes, our exascale program includes five main components. Not just hardware, not just software – all of it. I will detail the five components. I think this is going to be the answer to your question. The first component is an open exascale platform, its code name SEQUANA and it will scale up to tens of thousands of nodes. It’s really the foundation for exascale for Bull. It will be able to integrate successive generations of different technologies, CPUs, accelerators, whatever becomes available for maximum performance. It will support state of the art, fast interconnects such as the Bull exascale interconnect. This takes me to the second component of our program.

insideHPC: What is that? Is that InfiniBand or what is it?

Pascale Bernier-Bruna: BXI, that’s the code name for our Bull Exascale Interconnect. It’s an HPC interconnect designed by Bull. We are aiming to free up CPU performance by getting rid of the communication overhead that takes up so much CPU time today.

insideHPC: So that’s the second piece of this? What is the third?

Pascale Bernier-Bruna: Third piece is the matching software environment, Bull supercomputer suite – it will be completely revamped of course to manage performance, data management of possibly millions of hardware and software elements. Of course, it’s an important element. The fourth element is a series of server – the first model is shown on our booth. It’s servers dedicated to– in memory databases so it’s the first part of the program that will be available, first model available next year. It’s a long way to the exascale.

insideHPC: Let me stop you there, because a lot of people don’t know that when we talk about exascale, it isn’t just exaFLOPS which is floating point operations, there is exaOPS which is about more scalar things as well, needs to be there to make this a useful device. So is that where you’re headed with that fourth component?

Pascale Bernier-Bruna: Yes, that’s it, that’s exactly it.

insideHPC: ExaOPS?

Pascale Bernier-Bruna: Yes. The fifth component – I’ve kept it for last but it’s really most important, is a set of services to help our customers optimize their obligation for exascale, because there is no point in reaching the exascale if you just don’t know how to efficiently use all this million way parallelism. As a conclusion, I would like to say that Bull has been in HPC for more than ten years now. We are now a major HPC player in Europe. Now that we are part of the Atos group, we are entering a new dimension, because we are going to be able to leverage the Atos worldwide presence. They have 86,000 employees throughout the world, so we’ll be able to leverage this to deliver exascale systems across the world, and of course to deliver less than exascale systems – lots of them across the world.

insideHPC: That’s exciting. For those that might know it, here’s how I see you guys, from my perspective as a reporter. I see Bull as the vendor that the French atomic agency, helped foster to create the machines they need, because they’re on the road to building the first fusion reactor. I think you guys will be first there, but that takes tremendous computation to make that happen. Now you’re part of Atos, it isn’t about just France anymore is it?

Pascale Bernier-Bruna: Yes. What you just said is it is part of the story of course and that’s how we got to HPC, that’s how we got started, and it’s a very important foundation, but it’s not the whole story. Now that we are an Atos company, it will be much easier for us to develop, especially geographically.

Sign up for our insideHPC Newsletter.