ARMdevices.net

Nvidia partners with Microsoft to provide Azure cloud NDv2 supercomputer instance that can use up to 800 Nvidia Tensor Core GPUs interconnected on a single Mellanox InfiniBand backend network. This enables customers to rent an entire AI supercomputer on-demand. Microsoft and Nvidia engineers used 64 NDv2 instances on a pre-release version of the cluster to train BERT, a popular natural language conversational AI model, in a mere three hours achieved in part with Nvidia’s CUDA-core technology and interconnects by Mellanox. This is an excerpt from the full keynote video which you can watch here: https://www.youtube.com/watch?v=69nEEpdEJzU