What is the Nvidia dgx

​The NVIDIA DGX is a line of servers and workstations built by NVIDIA that specialize in using GPGPU to accelerate deep learning applications[1]. The DGX platform is built from the ground up for enterprise AI, combining the best of NVIDIA software, infrastructure, and expertise[2]. The DGX system is designed to maximize data science productivity with enterprise-grade support from AI infrastructure and data center[3]. The typical design of a DGX system is based upon a rackmount chassis with a motherboard that carries high-performance x86 server CPUs, typically Intel Xeons, with GPU modules integrated into the system using a version of the SXM socket[1]. The DGX-1 is an integrated deep learning workstation that provides GPU computing power of 1 PetaFLOPS and is built on a hardware architecture with several key components designed for large-scale deep learning workloads[4]. The DGX-2 is the successor of the DGX-1 and uses sixteen Volta-based V100 32GB cards in a single unit, delivering 2 Petaflops with 512GB of shared memory for tackling massive workloads[1]. The NVIDIA DGX A100 is the world's first 5-petaflops system[5].

Citations:

[1] https://en.wikipedia.org/wiki/Nvidia_DGX

[2] https://www.nvidia.com/en-us/data-center/dgx-platform/

[3] https://www.nvidia.com/en-in/data-center/dgx-systems/

[4] https://www.run.ai/guides/nvidia-a100/nvidia-dgx

[5] https://youtube.com/watch?v=MY7jZGZw9vA