NVIDIA Announces DGX GH200 AI Supercomputer

Hello!!

NVIDIA announced the DGX GH200 AI supercomputer, which combines GH200 Grace Hopper superchips and the NVLink switching system to develop giant AI models for various applications. The supercomputer offers exaflop performance and a shared memory of 144 terabytes, revolutionizing the AI ​​infrastructure and allowing significant advances in this area.

The supercomputer is powered by NVIDIA GH200 Grace Hopper superchips and NVIDIA NVLink switching system.
- The goal is to develop giant next-generation models for generative AI language applications, recommendation systems and data analysis.
- The DGX GH200 features massive shared memory that combines 256 GH200 superchips to provide 1 exaflop of performance and 144 terabytes of shared memory.
- GH200 superchips eliminate the need for traditional CPU-GPU connection, offering higher bandwidth between GPU and CPU and lower power consumption.
- The DGX GH200 is the first supercomputer to combine Grace Hopper superchips with the NVIDIA NVLink switching system, allowing all GPUs to work together as a single GPU.
- The supercomputer offers 48 times more NVLink bandwidth than the previous generation, providing the power of a massive AI supercomputer with the simplicity of programming a single GPU.
- Google Cloud, Meta and Microsoft will be the first to have access to the DGX GH200 to explore its capabilities in generative AI workloads.
- NVIDIA also intends to make the DGX GH200 design available as a template for cloud service providers to customize for their infrastructure.

#NVIDIA #SupercomputadorDGXGH200AI #GH200GraceHopper #NVLink #IA #ArtificialIntelligence #GiantModels #AILanguage #RecommendationSystems #DataAnalysis #Exaflop #SharedMemory #SuperchipsGH200 #CPU-GPU #Bandwidth #PowerConsumption #GoogleCloud #Meta #Microsoft #Loads DeLabor #IAInfrastructure #CloudServiceProviders #IAAdvances