Nvidia announced a new class of large-memory AI supercomputer, Nvidia DGX, that uses NVIDIA GH200 Grace Hopper Superchips and the NVIDIA NVLink Switch System power. The chipmaker also announced an AI-enabled Avatar Cloud Engine (ACE) for games and the production of its GH200 Grace Hopper Superchips.
The AI supercomputer, DGX GH200, is designed 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 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, the company shared in a release.
The DGX GH200 is the first supercomputer to pair Grace Hopper Superchips with the NVIDIA NVLink Switch System, a new interconnect that enables all GPUs in a DGX GH200 system to work together as one. The previous generation system only provided eight GPUs to be combined with NVLink as one GPU without compromising performance.
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
The DGX GH200 architecture provides 48x more NVLink bandwidth than the previous generation, the company claims.
Avatar Cloud Engine (ACE) for games
Nvidia also announced the Avatar Cloud Engine (ACE), a custom AI model to bring customised speech, conversation, and animation models into software and games.
The Ace AI model enables developers of middleware, tools, and games to build and deploy AI-enabled conversations with non-playable characters in games.
Building on Nvidia Omniverse, ACE for game developers optimises AI foundation models for speech, conversation, and character animation, including Nvidia NeMo, Riva, and Omniverse audio.
While Nvidia NeMo, a large language model, can be customized with lore and character backstories, Riva can be utilised for automatic speech recognition and text-to-speech to enable live speech conversation.
Developers can also use Nvidia Omiverse Audio2Face to create expressive facial animation of a game character to match any speech track, the company said.
Developers can integrate the entire NVIDIA ACE for Games solution or use only the components they need.
GH200 Grace Hopper Superchip
Nvidia announced its GH200 Grace Hopper Superchip is in full production. The GH200-powered systems join more than 400 system configurations powered by different combinations of Nvidia’s latest CPU, GPU, and DPU architectures created to help meet the surging demand for generative AI.
The GH200 delivers up to 900GB/s total bandwidth 7x higher bandwidth than the standard PCIe Gen5 lanes found in traditional accelerated systems.
The coming portfolio of systems accelerated by Nvidia Grace, Hopper, and Ada Lovelace architectures will provide broad support for the Nvidia software stack, which includes Nvidia AI, the Nvidia Omniverse platform, and Nvidia RTX technology.
“Generative AI, large language models, and recommender systems are the digital engines of the modern economy. DGX GH200 AI supercomputers integrate NVIDIA’s most advanced accelerated computing and networking technologies to expand the frontier of AI”, Jensen Huang, founder, and CEO of Nvidia, said.