In terms of OS, Spark is an ARM-based system that runs Nvidia. Dgx duckan Ubuntu Linux-based operating system designed specifically for GPU processing. It comes preloaded with Nvidia's AI software stack, including the company's CUDA libraries and NIM microservices.
Prices for the DGX Spark start at $3,999. This may seem like a lot, but considering the cost of high-end GPUs with large amounts of VRAM, such as the RTX Pro 6000 (around $9,000), or AI server GPUs (around $25,000 for the base level). H100), the DGX Spark may represent a much less expensive option overall, although it's nowhere near as powerful.
Actually, in accordance with According to Register, the GPU processing performance of the GB10 chip is roughly equivalent to the RTX 5070. However, the 5070 is limited to 12GB of video memory, which limits the size of AI models that can be run on such a system. With 128GB of unified memory, the DGX Spark can handle much larger models, albeit at slower speeds than, say, the RTX 5090 (which typically comes with 24GB of RAM). For example, to run a larger version of OpenAI containing 120 billion parameters, gpt-oses language model, you'll need about 80GB of memory, which is much more than you can get in a consumer GPU.
Call back to 2016
Nvidia founder and CEO Jensen Huang celebrated the DGX Spark launch event by personally delivering one of the first devices to Elon Musk at SpaceX's Starbase facility in Texas, echoing similar delivery Huang approached Musk at OpenAI in 2016.
“In 2016, we built DGX-1 to give artificial intelligence researchers their own supercomputer. The first system I gave to Elon from a small startup called OpenAI, and from that came ChatGPT,” Huang said in a statement. “DGX-1 ushered in the era of artificial intelligence supercomputing and ushered in the scaling laws that drive modern artificial intelligence. With DGX Spark we are returning to this mission.”
