ailocalinference

← Glossary

DGX Station

NVIDIA's desk-side workstation line — a data-center GPU under your desk

Hardware

In plain English

A computer the size of a small tower that holds the kind of AI chip normally found in a server room, so it can sit in an office instead.

Technically

NVIDIA's workstation-class systems built around data-center silicon rather than consumer GeForce parts, with large coherent memory, server-grade interconnect and the full CUDA and NGC software stack preinstalled. The line is positioned above the compact GB10 machines (DGX Spark and its partner variants) and below rack-mounted DGX servers. For local inference the draw is memory capacity and bandwidth in a box that runs on a standard office circuit; the cost is price, and power and heat well above a single desktop GPU.

Related terms

DGX SparkMikroTik CRS812-8C-4XQ-IN

Lessons that use it

More in Hardware

DGX SparkGB10 SuperchipMikroTik CRS504-4XQ-INMikroTik CRS804-4C-INMikroTik CRS812-8C-4XQ-INNVLink-C2CRoCEv2 (RDMA over Converged Ethernet)VRAM