NVIDIA
NVIDIA Grace Hopper Ushers in a New Era of AI Supercomputing ……
Energy-efficient Garce-powered systems, including Switzerland’s Alps supercomputer and France’s EXA1-HE supercomputer, deliver 200 exaflops of AI performance and lead groundbreaking research from climate and weather to scientific exploration
[Image: https://prtimes.jp/i/12662/463/resize/d12662-463-88ab72c061bdb54885c8-0.jpg&s3=12662-463-5bda8cc25a6f45fe0eceaf23d96fafe8-1600×900.jpg] Hamburg, Germany, May 12, 2024 – ISC – To drive a fundamental transition to AI-powered systems in the high-performance computing industry, NVIDIA today announced that nine new supercomputers worldwide will be powered by NVIDIA Grace Hopper ( TM) Superchip (https://www.nvidia.com/ja-jp/data-center/grace-hopper-superchip/) to accelerate scientific research and discovery. These systems deliver a total of 200 exaflops (2 operations per second) of energy-efficient AI processing power.
New supercomputers powered by Grace Hopper on the horizon include the EXA1-HE in France, built by CEA and Eviden, and the EXA1-HE in Poland at the Cyphronet Scientific Computing Center, built by Hewlett-Packard Enterprise (HPE). Helios, located at the Swiss National Supercomputing Center and built by HPE; JUPITER at the Jülich Supercomputing Center in Germany; DeltaAI at the Supercomputer Applications Laboratory at the University of Illinois at Urbana-Champaign; and the Center for Computational Science at the University of Tsukuba and Tokyo. These include Miyabi, a state-of-the-art joint HPC infrastructure facility (JCAHPC) jointly operated by the University Information Technology Center.
In April, France’s CEA (Atomic and Alternative Energy Agency) and Eviden, a company of the Atos Group, announced the operation of an EXA1-HE supercomputer based on Eviden’s BullSequana XH3000 technology. The BullSequana XH3000 architecture features a new patented hot water cooling system, and EXA1-HE features 477 compute nodes based on Grace Hopper.
NVIDIA’s Vice President of Hyperscale and HPC Ian Buck said: “AI is poised to accelerate research on climate change, accelerate drug discovery and development, and deliver breakthroughs in many other areas. It is becoming an indispensable part of HPC.”
In addition to the above, Isambard-AI and Isambard 3 at the University of Bristol in the UK, as well as systems at Los Alamos National Laboratory and the Texas Advanced Computing Center in the US, are also developing NVIDIA Arm-based supercomputers using the Grace CPU Superchip and Grace Hopper platforms. We’re about to join the growing wave.
Sovereign AI
The trend to build new, more energy-efficient, AI-powered
supercomputers continues to accelerate, with countries around the world using sovereign AI (https://blogs.nvidia.co.jp/blog/what- is-sovereign-ai/) and seeks to foster innovation by investing in domestically owned and hosted data, infrastructure and workforce. Combining the Arm-based NVIDIA Grace CPU and Hopper GPU architecture with NVIDIA NVLink(R)-C2C interconnect technology, the GH200 serves as the engine powering supercomputing centers around the world. Many centers are planning to move from system installation to actual scientific research in months rather than years.
Phase 1 of Isambard-AI consists of an HPE Cray Supercomputing EX2500 with 168 NVIDIA GH200 Superchips, making it the most efficient supercomputer to date. When the remaining 5,280 NVIDIA Grace Hopper Superchips are installed at the University of Bristol’s National Composites Center this summer, they will increase performance by approximately 32 times.
Professor Simon McIntosh-Smith from the University of Bristol said: “Isambard-AI will enable the UK to become a global leader in AI and foster open science innovation both at home and abroad. Our
collaboration with NVIDIA will enable us to deliver phase one of this project in record time. When completed this summer, it will
dramatically increase performance and advance data analysis, drug discovery, climate research, and many other fields.”
Accelerate scientific discovery
NVIDIA’s accelerated computing platform includes GPUs based on the NVIDIA Hopper(TM) architecture
(https://www.nvidia.com/ja-jp/data-center/h100/), NVIDIA Grace CPU Superchip, and NVIDIA Grace Consists of Hopper Superchip, NVIDIA Quantum-2 InfiniBand
(https://www.nvidia.com/ja-jp/networking/quantum2/) networking and the full suite of NVIDIA AI and HPC software.
About NVIDIA
Since its founding in 1993, NVIDIA (https://www.nvidia.com/ja-jp/) (NASDAQ: NVDA) has been a pioneer in accelerated computing. Invented by the company in 1999, the GPU is fueling the growth of the PC gaming market, redefining computer graphics, and igniting the modern AI era while helping to digitize industries. NVIDIA is a full-stack computing company with data center-scale products that are currently reshaping the industry. For more information, follow this link:
https://nvidianews.nvidia.com/
More details about this release:
https://prtimes.jp/main/html/rd/p/000000463.000012662.html