Overview
The NVIDIA DGX™ Spark is a next-generation AI workstation designed to meet the increasing demands of modern AI development. Built from the ground up for AI workloads, the DGX Spark is designed to handle everything from training to fine-tuning and inference for large AI models. Whether you're a researcher, data scientist, or developer, the DGX Spark is built to enable you to accelerate your AI projects locally.
NVIDIA DGX Spark
- Built on NVIDIA GB10 Grace Blackwell Superchip
- Up to 1000 TOPS of AI performance using FP4
- 128 GB of coherent, unified system memory
- Support for up to 200 billion parameter models
DGX Spark Technical Specifications
| Specification | Details |
|---|---|
| Architecture | NVIDIA Grace Blackwell |
| GPU | NVIDIA Blackwell Architecture |
| CPU | 20-core Arm, 10 Cortex-X925 + 10 Cortex-A725 Arm |
| CUDA Cores | NVIDIA 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 M.2 with self-encryption |
| USB | 4x USB Type-C |
| 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 | TBD |
| Display Connectors | 1x HDMI 2.1a |
| NVENC / NVDEC | 1x / 1x |
| OS | NVIDIA DGX™ OS |
| System Dimensions | 150 mm x 150 mm x 50.5 mm (5.91 in x 5.91 in x 1.99 in) |
| System Weight | 1.2 kg |
Key Features at a Glance
| Feature | Details |
|---|---|
| Next-Generation Hardware | Powered by the GB10 Grace Blackwell Superchip with 1,000 AI TOPS Blackwell GPU and 20-core Arm CPU. Ideal for demanding AI workloads. |
| Unified System Memory | 128GB of coherent memory shared by the CPU and GPU, enabling seamless access to large AI models. |
| Flexible AI Deployment | Configure as a personal AI cloud or connect to existing systems for scalable solutions. |
| Compact and Power Efficient | Easily fits on any desk and powered by a standard wall outlet. |
| Scalable Performance | Link up to two DGX Spark systems via ConnectX-7 for working with models up to 405B parameters. |
DGX Spark Software Stack
The DGX Spark comes pre-configured with NVIDIA's enterprise-grade software stack, designed to simplify the development and deployment of AI models:
- NVIDIA AI Enterprise: A suite of AI tools for the most demanding workloads.
- NVIDIA DGX OS: Optimized for AI workloads and ease of use.
- Supported Frameworks: TensorFlow, PyTorch, and more.
AI Model Support
The DGX Spark is optimized for a wide range of AI frameworks and tools. It supports:
- TensorFlow
- PyTorch
- MATLAB
- Holoscan
- Riva
- Metropolis
- Isaac
- NVAIE
DGX Spark allows developers to run AI models locally, including those too large to run on traditional laptops and desktops.
Scalable AI Solutions
For those who need more power, the DGX Spark allows users to scale their AI workloads by clustering two units together via ConnectX-7 for up to 405 billion parameters. Whether you're working on complex deep learning models or large-scale AI training, DGX Spark offers a compact, power-efficient solution with the ability to grow as your needs evolve.
Unlock the power of AI and revolutionize your AI workflow with the DGX Spark.