
NVIDIA DGX™ Spark : Maximum AI performance in a compact form factor
The NVIDIA DGX Spark is a compact yet highly powerful AI desktop supercomputer, utilizing the NVIDIA GB10 Grace Blackwell Superchip designed specifically for AI applications. Featuring the NVIDIA Grace Blackwell architecture, the DGX Spark delivers AI processing capability up to 1,000 TOPS (FP4), making it ideal for AI developers, researchers, data scientists, and students who want to build, fine-tune, and deploy large AI models.
Main features of NVIDIA DGX™ Spark
Processor chip: NVIDIA GB10 Grace Blackwell Superchip, including a Blackwell architecture GPU with 5th generation Tensor Cores and 4th generation RT Cores, and a 20-core CPU (10 Cortex-X925 + 10 Cortex-A725 Arm).
Memory: 128GB LPDDR5x RAM (Unified System Memory) with memory bandwidth up to 273 GB/s.
Storage: 1TB or 4TB NVMe M.2 SSD with self-encryption function.
Connectivity: HDMI 2.1a port, RJ-45 10GbE port, WiFi 7, Bluetooth 5.3, and 4 USB4 ports.
Operation: Runs on NVIDIA DGX OS, developed from Ubuntu 22.04, supporting large AI models such as NVIDIA Cosmos Reason and NVIDIA GR00T N1.





Accelerate all AI workloads. NVIDIA DGX Spark delivers the power of an AI supercomputer in a desktop-sized form, making it ideal for the workloads of AI developers, researchers, and data scientists.
Fine-tuning: Customize AI models with up to 70 billion parameters. Enhance the performance of pretrained models with fine-tuning on the NVIDIA DGX Spark, featuring a total system memory of 128GB. Fine-tune models with up to 70 billion parameters to tailor AI models and solutions to specific needs and use cases.
Data Science accelerates end-to-end data science workflows with NVIDIA RAPIDS™, from data preparation to model training, inference, and deployment. NVIDIA RAPIDS on NVIDIA DGX Spark simplifies development with accelerators that require no code changes and familiar APIs to speed up existing workloads or rapidly develop specific use-case solutions, easily scaling out to data centers or the cloud.
Edge Applications: Develop Edge applications using NVIDIA AI frameworks including Isaac™, Metropolis, and many more. DGX Spark provides an excellent platform for developing robotics, smart cities, and computer vision solutions. NVIDIA frameworks like Isaac, Metropolis, and Holoscan enable developers to leverage the power of NVIDIA DGX Spark to rapidly build Edge applications.
Performance and capabilities
DGX Spark offers very high AI processing power, capable of up to 1,000 TOPS (FP4), making it suitable for fine-tuning and inference of complex AI models. Additionally, it supports inter-machine connectivity for larger models using NVIDIA ConnectX Networking technology.
Specifications of NVIDIA DGX™ Spark
Architecture | NVIDIA Grace Blackwell |
GPU | Blackwell Architecture |
CPU | 20 core Arm, 10 Cortex-X925 + 10 Cortex-A725 Arm |
CUDA Cores | Blackwell Generation |
Tensor Cores | 5th Generation |
RT Cores | 4th Generation |
Tensor Performance1 | 1000 AI TOPS |
System Memory | 128 GB LPDDR5x, unified system memory |
Memory Interface | 256-bit |
Memory Bandwidth | 273 GB/s |
Storage | 1 or 4 TB NVME.M2 with self-encryption |
USB | 4x USB 4 TypeC (up to 40Gb/s) |
Ethernet | 1x RJ-45 connector 10 GbE |
NIC | ConnectX-7 Smart NIC |
Wi-Fi | WiFi 7 |
Bluetooth | BT 5.3 |
Audio-output | HDMI multichannel audio output |
Power Consumption | 170W |
Display Connectors | 1x HDMI 2.1a |
NVENC | NVDEC | 1x | 1x |
OS | NVIDIA DGX™ OS |
System Dimensions | 150 mm L x 150 mm W x 50.5 mm H |
System Weight | 1.2 kg |
Summary
NVIDIA DGX Spark is a powerful tool for AI development in a compact desktop form, featuring high performance and impressive AI processing capabilities, enabling AI developers to efficiently build and fine-tune AI models.
Read more product information at: (https://mtxr.co/shop/nvidia-dgx-spark/)
For inquiries, contact LINE ID: @metaxr ( https://lin.ee/MVuGWFC )