Atos Launches First Supercomputer Equipped with NVIDIA A100 GPU

New Atos accelerator blade allows JUWELS BullSequana method to turn into the most highly effective supercomputer in Europe

Paris, France, May well fourteen 2020

Atos, a worldwide chief in electronic transformation, these days announces its new BullSequana X2415, the initially supercomputer in Europe to combine NVIDIA’s Ampere subsequent-era graphics processing device architecture, the NVIDIA® A100 Tensor Main GPU. This new supercomputer blade will provide unprecedented computing electric power to strengthen application overall performance for HPC and AI workloads, tackling the issues of the exascale period. The BullSequana X2415 blade will boost computing electric power by a lot more than 2* and improve vitality consumption thanks to Atos’ 100{744e41c82c0a3fcc278dda80181a967fddc35ccb056a7a316bb3300c6fc50654} remarkably successful h2o-cooled patented DLC (Immediate Liquid Cooling) remedy, which employs warm h2o to great the machine.

Forschungszentrum Julich Ralf-Uwe Limbach

Forschungszentrum Jülich will combine this new blade into its booster module, extending its present JUWELS BullSequana supercomputer, earning it the initially method around the globe the use this new engineering. The JUWELS Booster will present researchers throughout Europe with considerably enhanced computational assets. Some of the initiatives it will gas are the European Commission’s  Human Mind Project  and the Jülich Laboratories of  “Climate Science” and “Molecular Systems”. After entirely deployed this summer time the upgraded supercomputing method, operated beneath ParTec’s application ParaStation Modulo, is predicted to present a computational peak overall performance of a lot more than 70 Petaflops/s earning it the most highly effective supercomputer in Europe and a showcase for European exascale architecture.

“By integrating the ideal systems obtainable, together with this latest blade from Atos which involves NVIDIA’s highly effective A100 GPU into our supercomputer, we are equipped to strengthen our supercomputing electric power. This is an vital milestone on our path to arrive at our supreme aim of giving exascale capabilities to science and field” stated Prof. Thomas Lippert, Director of the Jülich Supercomputing Centre (JSC).

“We’re very pleased to be one of the initially supercomputer makers around the globe to contain NVIDIA’s A100 GPU and to be equipped to offer you researchers around the globe the greatest application overall performance for their HPC and AI workloads” stated Agnès Boudot, Senior Vice President, Head of HPC & Quantum at Atos.  “We’re genuinely thrilled to be working with Jülich to strengthen its computing electric power and to empower its JUWELS method to turn into the most highly effective supercomputer in Europe.”

“Atos is one of the top rated players entire world-vast in the high finish of the server current market. It is the leading European-based mostly vendor of HPC systems, mostly due to the overall performance and electric power of its BullSequana systems, its powerful technological abilities and innovation and its potential to effectively deal with significant-scale initiatives globally. It has doubled its existence in the TOP500 position about the previous five years and its server current market share around the globe carries on to increase. We see Atos as one of the essential supercomputing makers that will likely reward about the subsequent couple years as desire for exascale systems turns into a lot more significant.” said Steve Conway, Senior Advisor at Hyperion Analysis.

The new blade options an NVIDIA HGX-A100 base-board geared up with 4 NVIDIA A100 GPUs connected by way of 3rd-era NVIDIA NVLink™ engineering, together with two AMD EPYC CPUs and up to 4 NVIDIA Mellanox InfiniBand ports connected by way of a Dragonfly+ configuration. The NVIDIA Ampere architecture is a groundbreaking engineering accomplishment that delivers a substantial overall performance improvement about the earlier era for AI coaching and inference. The NVIDIA A100 Tensor Main GPU options the world’s largest seven-nanometer processor, with a lot more than 54 billion transistors. This workforce of systems, alongside with the NVIDIA current market-leading ecosystem of about 700 GPU-accelerated HPC programs and support from NGC, will enable researchers to hit the ground managing with containerized application optimized for the advancement and deployment of GPU-accelerated HPC and AI initiatives.

The BullSequana X2415 will be obtainable in Q2 of 2020.

###

* in contrast to Atos’ present GPU blade

X2415 specs

 1U kind element blade with 1 acceleration node.

Immediate Liquid Cooling.

For each node:

1 x CPU Board with

• 2 x AMD EPYC/Milan CPUs
• sixteen x DDR4 32GB @3200MTs minimal

1 x GPU board “HGX-A100” with 4 x NVIDIA A100 GPUs

2 x Interconnect Mezzanine boards (BXI or HDR200)

Optional 1 x M.2 SATA SSD up to 1.92TB or 1 x M.2 NVMe SSD 960GB

About Atos

Atos is a worldwide chief in electronic transformation with a hundred and ten,000 workforce in 73 nations around the world and annual profits of € 12 billion. European selection one in Cloud, Cybersecurity and Higher-Overall performance Computing, the Team offers finish-to-finish Orchestrated Hybrid Cloud, Large Facts, Business enterprise Programs and Electronic Workplace remedies. The Team is the All over the world Data Engineering Partner for the Olympic & Paralympic Games and operates beneath the models Atos, Atos|Syntel, and Unify. Atos is a SE (Societas Europaea), mentioned on the CAC40 Paris inventory index.

The function of Atos is to support structure the potential of the facts place. Its abilities and expert services support the advancement of knowledge, education and learning and investigate in a multicultural solution and add to the advancement of scientific and technological excellence. Throughout the entire world, the Team allows its prospects and workforce, and members of societies at significant to are living, function and create sustainably, in a risk-free and protected facts place.

Push get in touch with:

Laura Fau | [email protected] | +33 six 73 64 04 18 | @laurajanefau