NVIDIA DGX Spark™
NVIDIA DGX Spark™ Photo credit: NVIDIA

Nvidia has begun shipping DGX Spark, which it describes as the world’s smallest AI supercomputer. Founder and CEO Jensen Huang hand-delivered one of the first units to Elon Musk at SpaceX in Texas.

The delivery echoes a moment in 2016 when Huang hand-delivered the first DGX-1 supercomputer to Musk at OpenAI, from which ChatGPT emerged. Huang said: “In 2016, we built DGX-1 to give AI researchers their own supercomputer. I hand-delivered the first system to Elon at a small startup called OpenAI — and from it came ChatGPT, kickstarting the AI revolution. DGX-1 launched the era of AI supercomputers and unlocked the scaling laws that drive modern AI. With DGX Spark, we return to that mission — placing an AI computer in the hands of every developer to ignite the next wave of breakthroughs.”

DGX Spark delivers one petaflop of AI performance and 128GB of unified memory in a compact desktop form factor, addressing the challenge that AI workloads are outgrowing the capabilities of standard PCs, workstations and laptops used by millions of developers. The system allows developers to run inference on AI models with up to 200 billion parameters and fine-tune models of up to 70 billion parameters locally.

Built on the Nvidia Grace Blackwell architecture, DGX Spark integrates GPUs, CPUs, networking, CUDA libraries and Nvidia AI software. The system is accelerated by an Nvidia GB10 Grace Blackwell Superchip, featuring ConnectX-7 200Gb/s networking and NVLink-C2C technology, which provides five times the bandwidth of fifth-generation PCIe with 128GB of CPU-GPU coherent memory.

Kyunghyun Cho, professor of computer and data science at the NYU Global Frontier Lab, says: “DGX Spark allows us to access peta-scale computing on our desktop. This new way to conduct AI research and development enables us to rapidly prototype and experiment with advanced AI algorithms and models — even for privacy- and security-sensitive applications, such as healthcare.”

Early recipients testing and optimising for DGX Spark include Anaconda, Cadence, ComfyUI, Docker, Google, Hugging Face, JetBrains, LM Studio, Meta, Microsoft, Ollama and Roboflow.

Acer, ASUS, Dell Technologies, GIGABYTE, HP, Lenovo and MSI are launching DGX Spark systems. Starting 15 October, DGX Spark can be ordered on Nvidia.com, with partner systems available from Micro Center stores in the US and Nvidia channel partners worldwide.

Leave a Reply

Your email address will not be published. Required fields are marked *

You May Also Like

Super-intelligent AI could ‘play dumb’ to trick evaluators and evade controls

The dream of an AI-integrated society could turn into a nightmare if…

Satellite dataset uses deep learning to map 9.2 million kilometres of roads

Researchers have combined deep-learning models with high-resolution satellite imagery to classify 9.2…

Universities quietly deploying GenAI to ‘game’ £2bn research funding system

UK universities are widely using generative AI to prepare submissions for the…

AI guardrails defeated by poetry as ‘smarter’ models prove most gullible

The world’s most advanced artificial intelligence systems are being easily manipulated into…

Researchers hijack X feed with ad blocker tech to cool political tempers

Scientists have successfully intercepted and reshaped live social media feeds using ad-blocker-style…

Doing good buys forgiveness as CSR becomes ‘insurance’ against layoffs

Companies planning to slash jobs or freeze pay should start saving the…