The use of high-performance computing (HPC) platforms has been instrumental in accelerating deep learning, which is a vital part of any AI – but can require specialized computing facilities.
To address this potential bottleneck the companies are introducing a joint initiative, the Fujitsu and NVidia GPU Cluster. Part of Fujitsu’s Zinrai initiative1, a collective framework of AI capabilities that are available to Fujitsu customers, the cluster makes HPC capabilities more easily available to start-ups, based on the power of Fujitsu’s Primergy server infrastructure and NVidia GPU’s.
The Fujitsu and NVidia GPU Cluster will be based at the Drahi X-Novation Center, the incubator of the prestigious Ecole Polytechnique in Paris-Saclay, which also hosts Fujitsu’s AI Center of Excellence (CoE)1. Intended to help underpin the development of a strong AI ecosystem in Europe, the GPU cluster offers a breakthrough opportunity for new startups or students in educational institutions.
The cluster is based on NVidia Tesla V100 Tensor Core GPUs and Fujitsu Primergy servers. Together, these provide the powerful computing capabilities needed to develop deep learning projects. The relevant compute capacity will be distributed based on the needs of individual AI projects being developed by startups, by Polytechnique students and by the Fujitsu AI CoE.
Axel Mery, chief technology officer of Fujitsu France and Head of Fujitsu’s AI Center of Excellence, says: “Deep learning at the heart of AI requires much more powerful compute resources than conventional computational power and limited access to these resources can be a brake on innovation.
“As part of our EUR 50-million investment over five years to support digital innovation in France, we are now making these often-unattainable computational resources available to our most innovative developers at exactly the time they need them most – regardless of their stage of funding. Through this initiative in Paris, as well as Fujitsu’s ongoing investment in its Zinrai² AI offering, we see huge potential to develop new technologies and human-centric AI solutions.”
Serge Palaric, vice-president: business development enterprise for South Europe & Embedded EMEA at NVidia, comments: “Through this initiative, students and startups can gain access to GPU clusters that accelerate the critical deep learning stage for any AI. Our work with Fujitsu and Ecole Polytechnique is another pillar in our support for the research, education and innovation ecosystem enabling the brightest minds in tech.”