NVIDIA DGX
Main Features
The DGX system features eight NVIDIA GPUs from a range of models from the A100, to the latest HG200. The GPUs are interconnected via NVIDIA NVSwitch, enabling high-speed data transfer between GPUs. The SuperPOD standard allows DGX modules to be quickly scaled into single data center clusters.
Technology and Features
- DGX systems can support a wide range of applications, including deep learning, data analytics and high-performance computing.
- In particular, the DGX H100, part of the DGX platform, utilizes eight NVIDIA H100 GPUs with a total memory of 640 GB and data transfer rates of up to 7.2 TB/s between GPUs.
.
NVIDIA HGX
.
NVIDIA HGX is a modular version of the DGX designed for OEMs to integrate NVIDIA GPU modules into their systems. This solution enables the creation of customizable systems with maximum performance based on NVIDIA technologies. HGX includes the same GPUs and technologies as DGX, but allows manufacturers to customize the rest of the system to their liking.
Applications of NVIDIA DGX and HGX
Both platforms are ideal for enterprise AI, including deep learning, big data analytics, and machine learning model building. They are widely used in a variety of industries, including healthcare, automotive, energy and technology, where powerful computing resources are required for innovative solutions and workflow optimization.
Previous article