0
Nvidia has announced its latest chip, the Nvidia H200, designed to power supercomputers for generative AI models. The chip is based on the Nvidia Hopper architecture and features advanced memory capabilities for handling large amounts of data. It is the first GPU to offer faster and larger memory with HBM3e, providing 141GB of memory at 4.8 terabytes per second. The Nvidia H200 will be available from system manufacturers and cloud service providers in the second quarter of 2024. Nvidia’s GPUs are known for their parallel processing architecture, making them ideal for accelerating generative AI models.
