Recently, Nvidia announced that it is collaborating with Microsoft to create a “massive” cloud computing focused on artificial intelligence (AI). The supercomputer to be installed will be hosted on Microsoft’s Azure platform.It is reported that tens of thousands of high-end Nvidia GPUs will be used for applications such as deep learning and large language models. The companies aim to make the new supercomputer one of the most powerful AI supercomputers in the world.
Multi-year partnership with Nvidia from Microsoft
The platform to be created will have thousands of units of the Hopper H100, which is arguably the world’s most powerful GPU, which Nvidia launched in October. Nvidia will also use thousands of its second most powerful GPU, the A100, in the installation. These hardware will use the Quantum-2 InfiniBand network platform and will achieve data transfer rates of 400 gigabits per second between servers.As part of its collaboration with Microsoft, Nvidia says it will use Azure virtual machine samples to investigate advances in artificial intelligence or self-learning algorithms that can generate text, code, images, video or audio. (like OpenAI’s GPT-3 or DALL-E 2) Meanwhile, Microsoft will optimize its DeepSpeed library for new Nvidia hardware, aimed at reducing computing power and memory usage during AI training workloads.
The companies didn’t elaborate on when the new supercomputer will be ready, but noted that the announcement marks the start of a “multi-year collaboration.” It also seems likely that the capacity of cloud computing will increase over time.
21460
