NVIDIA DGX Station A100 Announced

by Adam Armstrong November 16, 2020

written by Adam Armstrong November 16, 2020

Today at SC20 NVIDIA announced its AI data-center-in-a-box with the NVIDIA DGX Station A100. This marks the second generation of the DGX and claims to be the world’s only petascale integrated AI workgroup server. This server comes with up to 320GB of GPU memory and can fit under a desk.

NVIDIA DGX Station A100

The NVIDIA DGX Station A100 has four of the newly announced NVIDIA A100 GPUs that are said to deliver up to 2.5 petaflops. The GPUs are interconnected with NVIDIA NVLink that will provide up to 320GB of memory ideal for memory hungry applications in AI and ata science. The NVIDIA DGX Station A100 also supports the company’s Multi-Instance GPU (MIG) technology. MIG allows a single DGX Station A100 provides up to 28 separate GPU instances to run parallel jobs and support multiple users without impacting system performance.

One of the neat facets about the NVIDIA DGX Station A100 is that it is incredibly powerful, yet small and doesn’t’ need the type of power and cooling that data center gear typically does. This means that it can be used in remote locations without a bunch of extra infrastructure being added.

Today the company also announced the new NVIDIA A100 80GB GPU. Adding the new 80GB version to the DGX A100 systems takes the memory capacity up to 640GB to tackle even larger datasets and models. NVIDIA states that the new NVIDIA DGX A100 640GB systems can also be integrated into the NVIDIA DGX SuperPOD Solution for Enterprise, allowing organizations to build, train and deploy massive AI models on turnkey AI supercomputers available in units of 20 DGX A100 systems.

Availability

NVIDIA DGX Station A100 and NVIDIA DGX A100 640GB systems will be available this quarter through NVIDIA Partner Network resellers worldwide.

NVIDIA

Engage with StorageReview