Nvidia brings CUDA-Q platform for hybrid computing to supercomputers in Germany, Japan and Poland

May 13. (Portaltic/EP) –

The technology of Nvidia advances ongoing scientific research globally with the deployment of Nvidia’s GH200 Grace Hopper ‘superchips’ in nine supercomputers and the adoption of the open source CUDA-Q hybrid computing platform in three high-performance computers.

The great scientific advancessuch as the discovery of new drugs, research on climate change or nuclear fusion energy, are developed in simulations that drive supercomputers and artificial intelligence (AI), but also quantum computers.

The technology company has shared a series of new developments in supercomputing and quantum computing that take advantage of the power of its processors, within the framework of ISC High Performance 2024, held in Hamburg (Germany).

This is the case of nine supercomputers around the world that have begun to use Grace Hopper chips, a next-generation platform that it detailed in November last year and that has been designed for the era of accelerated computing and generative AI.

This Grace Hopper GH200 ‘superchip’ combines the HGX H200 GPU with Hopper architecture with an Nvidia Grace CPU to offer computing power of 200 exaflops of AI performance.

Additionally, Nvidia has announced that supercomputers from Germany, Japan and Poland will implement the open source platform CUDA-Qpresented in July 2022 and designed for the hybrid classical-quantum computing in high-performance computers, as reported in the specialized media Network World and Venture Beat.