file-en/CustomerStory_nvidia_Letter_web.pdf
As an institute of the Bavarian Academy of Sciences and Humanities, the Leibniz Supercomputing Centre (LRZ) is the high-performance computing center for Munich’s top universities and research institutions in Germany and Europe. Altair provides and installs NVIDIA’s DGX systems and infrastructure at the LRZ and at customer sites. To meet the special needs of best-in-class AI and ML research at the Munich Center for Machine Learning (MCML), the Altair team had to design a cluster solution that provides strong nodes and big capacity to train a multitude of networks, and is able to develop highly scalable ML and AI applications. Thanks to Altair’s expertise in the parallelization of GPU applications the team was able to develop, prepare, and install a highly scalable system consisting of eight DGX A100 systems combined to a powerful cluster called DGX POD, while ensuring continuous power supply and the necessary cooling of the cluster.