The machine is powered by the new Nvidia GB10 Grace Blackwell Superchip, which offers a petaflop of AI computing performance, enabling the development, fine-tuning, and running of large AI models.
Nvidia co-founder Jensen Huang said: “AI will be mainstream in every application for every industry. With Project DIGITS, the Grace Blackwell Superchip comes to millions of developers. Placing an AI supercomputer on the desks of every data scientist, AI researcher and student empowers them to engage and shape the age of AI.”
The GB10 Superchip, a system-on-a-chip (SoC) based on the Nvidia Grace Blackwell architecture, delivers up to 1 petaflop of AI performance at FP4 precision. The Superchip integrates an Nvidia Blackwell GPU with the latest Cuda cores and fifth-generation Tensor Cores, connected via NVLink-C2C to a high-performance Nvidia Grace CPU, featuring 20 Arm-based power-efficient cores. MediaTek has contributed to the design.
Project DIGITS enables users to develop and run inference on models using their desktop systems, and then seamlessly deploy the models on accelerated cloud or data center infrastructure. Each unit comes with 128GB of unified, coherent memory, up to 4TB of NVMe storage, and the capability to link two supercomputers using Nvidia ConnectX® networking to run up to 405-billion-parameter models.
With the Grace Blackwell architecture, enterprises and researchers can prototype, fine-tune, and test models on local Project DIGITS systems running Linux-based Nvidia DGX OS, before deploying them on Nvidia DGX Cloud or data center infrastructure. The Nvidia AI Enterprise software platform provides the consistency needed across these environments.
Users have access to an extensive library of Nvidia AI software, including development kits, orchestration tools, and frameworks available in the Nvidia NGC catalogue and on the Nvidia Developer portal. This includes the Nvidia NeMo framework, Nvidia RAPIDS libraries, and common frameworks like PyTorch, Python, and Jupyter notebooks.
Project DIGITS will be available in May, starting at $3,000 from Nvidia and top partners.
