NVIDIA GH200 GPU Debuts On The Grace Hopper Superchips, New HGX Servers Revealed Too

NVIDIA's GH200 Hopper GPU has just been revealed for the Grace Hopper Superchips and will be powering HGX systems.

Press Release: NVIDIA today announced that the NVIDIA GH200 Grace Hopper Superchip is in full production, set to power systems coming online worldwide to run complex AI and HPC workloads.

The GH200-powered systems join more than 400 system configurations powered by different combinations of NVIDIA’s latest CPU, GPU, and DPU architectures including Grace, Hopper, Ada Lovelace, and BlueField — created to help meet the surging demand for generative AI.

At COMPUTEX, NVIDIA founder and CEO Jensen Huang revealed new systems, partners, and additional details surrounding the GH200 Grace Hopper Superchip, which brings together the Arm-based Grace CPU and Hopper GPU architectures using NVLink-C2C interconnect technology. This delivers up to 900GB/s total bandwidth — 7x higher bandwidth than the standard PCIe Gen5 lanes found in traditional accelerated systems, providing incredible compute capability to address the most demanding generative AI and HPC applications.

Global hyper scalers and supercomputing centers in Europe and the U.S. are among several customers that will have access to GH200-powered systems.

Full-Stack Computing Across Accelerated Systems

The coming portfolio of systems accelerated by the  Grace, Hopper, and Ada Lovelace architectures provides broad support for the NVIDIA software stack, which includes NVIDIA AI, the Omniverse platform, and RTX technology.

NVIDIA AI Enterprise, the software layer of the NVIDIA AI platform, offers over 100 frameworks, pre-trained models, and development tools to streamline the development and deployment of production AI, including generative AI, computer vision, and speech AI.

The Omniverse development platform for building and operating metaverse applications enable individuals and teams to work across multiple software suites and collaborate in real-time in a shared environment. The platform is based on the Universal Scene Description framework, an open, extensible 3D language for virtual worlds.

NVIDIA today announced a new class of large-memory AI supercomputer — an NVIDIA DGX supercomputer powered by NVIDIA® GH200 Grace Hopper Superchips and the NVIDIA NVLink Switch System — created to enable the development of giant, next-generation models for generative AI language applications, recommender systems, and data analytics workloads.

The NVIDIA DGX GH200’s massive shared memory space uses NVLink interconnect technology with the NVLink Switch System to combine 256 GH200 Superchips, allowing them to perform as a single GPU. This provides 1 exaflop of performance and 144 terabytes of shared memory — nearly 500x more memory than the previous generation NVIDIA DGX A100, which was introduced in 2020.

The NVIDIA RTX platform fuses ray tracing, deep learning, and rasterization to fundamentally transform the creative process for content creators and developers with support for industry-leading tools and APIs. Applications built on the RTX platform bring the power of real-time photorealistic rendering and AI-enhanced graphics, video, and image processing to enable millions of designers and artists to create their best work.

Availability

Systems with NVIDIA GH200 Grace Hopper Superchips are expected to be available beginning later this year. The DGX GH200 supercomputers are expected to be available by the end of the year.