NVIDIA Unveils Hopper GPU Architecture and Grace Superchips at GTC 2022
NVIDIA introduces the new Hopper GPU architecture to replace Ampere, alongside the debut of H100 GPUs and Grace CPU superchips designed for advanced AI and computing workloads.
NVIDIA reveals its next-generation Hopper GPU architecture at GTC 2022, officially marking the end of the Ampere era. This new architectural design focuses heavily on accelerating artificial intelligence and high-performance computing workloads. The shift represents a major leap forward in how graphics processing units handle complex data center operations.
The flagship H100 GPU makes its debut as the first product built on the Hopper architecture. It features significant performance upgrades tailored specifically for large language models and deep learning tasks. By optimizing for these demanding applications, NVIDIA aims to set a new industry standard for data center efficiency and processing speed.
In addition to the new graphics processors, NVIDIA introduces Grace CPU superchips that combine Arm-based processors with advanced memory capabilities. These superchips work in tandem with the Hopper GPUs to create incredibly powerful computing systems. Together, these unveilings give NVIDIA a comprehensive and highly competitive portfolio for the evolving enterprise technology landscape.