Atos Previews Energy-Efficient, AI-Augmented Hybrid Supercomputer

Stepping deeper into the period of exascale AI, Atos gave the first search at its future-generation substantial-general performance laptop or computer.

The BullSequana XH3000 brings together Atos’ patented fourth-era liquid-cooled HPC structure with NVIDIA technologies to supply both equally much more overall performance and strength performance.

Supplying users a choice of Arm or x86 computing architectures, it will appear in versions applying NVIDIA Grace, Intel, AMD or SiPearl processors. For accelerated computing, it supports nodes with four NVIDIA Tensor Core GPUs.

The XH3000 also flexibly employs network alternatives which include NVIDIA Quantum-2 InfiniBand and NVIDIA ConnectX-seven InfiniBand and Ethernet adapters to scale these effective computing nodes to HPC units capable of 10 mixed precision AI exaflops.

Hybrid HPC AI Quantum Program

The result is a flexible, hybrid computing platform able of operating the most demanding HPC simulations, AI employment and even rising workloads in quantum computing.

The BullSequana XH3000 “will no doubt empower, through the gateway of exascale, some of the essential scientific and industrial innovation breakthroughs of the potential,” explained Rodolphe Belmer, CEO of Atos, in a virtual event revealing the method.

With shoppers in more than 70 nations, Atos is #one in supercomputing in Europe, India and South America and particularly renowned in France, the place it maintains its headquarters as effectively as a production and R&D foundation.

A Wide Collaboration

Optimizations for the XH3000 had been between the first jobs for EXAIL, the joint Excellence AI Lab that Atos and NVIDIA declared in November.

John Josephakis, world wide vice president of sales and organization advancement for HPC/supercomputing at NVIDIA, congratulated the staff driving the process in a video information.

“By combining the nicely-known experience Atos has with NVIDIA AI and HPC technologies and work at our joint lab, this platform will make it possible for researchers to get substantial insights a lot speedier to grand challenges the two in supercomputing and industrial HPC,” he stated.

EXAIL’s perform spans local climate study, healthcare and genomics, quantum computing, edge AI/computer eyesight and cybersecurity. Its scientists can accessibility application frameworks this sort of as NVIDIA Clara for health care and NVIDIA Morpheus for stability as nicely as the NVIDIA cuQuantum SDK for quantum computing and the NVIDIA HPC SDK that operates hundreds of scientific and complex purposes.

A Extensive, Productive Connection

Atos designed one of Europe’s initial supercomputers to utilize the NVIDIA Ampere architecture, the JUWELS Booster at the Jülich Supercomputing Middle. It employs 3,744 NVIDIA A100 Tensor Main GPUs to produce two.five exaflops of blended-precision AI performance.

To deliver a deeper knowledge of weather adjust, Atos and NVIDIA scientists will run AI designs on the process, presently ranked No. 8 on the Leading500 checklist of the world’s swiftest supercomputers. Julich scientists used the process in April to carry out a state-of-the-artwork quantum circuit simulation.

Previous calendar year, Atos led deployment of BerzeLiUs, a process created on the NVIDIA DGX SuperPOD and Sweden’s largest supercomputer. The enterprise also has shipped supercomputing infrastructure in Europe, India and South The us primarily based on NVIDIA DGX units.

Up coming up, Atos is building Leonardo, a supercomputer at the Italian inter-university consortium CINECA. It will pack 14,000 NVIDIA A100 GPUs on an NVIDIA Quantum InfiniBand network and is envisioned to turn into the world’s speediest AI supercomputer, able of 10 exaflops of mixed-precision AI effectiveness.

With the 1st glimpse of the BullSequana XH3000, it is crystal clear there is substantially a lot more to come from the collaboration of Atos and NVIDIA.