NVIDIA DGX Spark

DGX personal AI computer, designed to build and run AI

  • A compact system for local AI development and commissioning

  • Grace Blackwell Superchip (GB10) -. integrated CPU + GPU

  • Up to 1 PFLOP (FP4) AI performance

  • 128 GB of unified memory -. working with large models

  • Supports models up to 200 billion parameters

  • Scaling to 405 billion parameters (NVIDIA ConnectX)

  • 4 TB NVMe - high-speed data storage

  • Format desktop

Product out of stock

Free shipping from €300

Promocja cenowa na model HDR-15-5

Product intended for professional use only
NVIDIA DGX Spark

NVIDIA DGX Spark

Description

 
DGX-class compact AI system

NVIDIA DGX Spark - local AI power in a desktop format

NVIDIA DGX Spark is a compact AI system designed for developers, researchers and data science teams who want to develop and run artificial intelligence models locally.

With its Grace Blackwell Superchip architecture and unified memory, the system enables large AI models to run with high performance and low latency.

DGX Spark enables the development and running of AI models of a scale previously reserved for data centre infrastructure.

Grace Blackwell Superchip

The system is based on the NVIDIA GB10 Grace Blackwell Superchip, which integrates the CPU and GPU in a single computing platform.

  • NVIDIA Blackwell GPU with Tensor Core 5. Generation
  • NVIDIA GraceCPU (20-core Arm architecture)

This approach allows for efficient data processing and optimised AI pipelines.

AI performance and memory

  • Up to 1 PFLOP (FP4) AI performance
  • 128 GB of unified memory shared by CPU and GPU

Unified memory simplifies large models and eliminates the need for manual data transfer management.

1 PFLOP
performance AI (FP4)
128 GB
unified memory
20
Arm CPU cores

Support for large AI models

DGX Spark enables large language models and generative systems to run locally.

  • models up to 200 billion parameters
  • scaling up to 405 billion parameters with NVIDIA ConnectX

Storage and integration

The system is equipped with 4 TB of NVMe storage for fast data storage and efficient operation with large data sets.

The compact, desktop form factor allows the system to be deployed directly into the AI team's working environment.

Technical Specification

Architecture NVIDIA Grace Blackwell
GPU NVIDIA Blackwell Architecture
CPU 20‑core Arm - 10× Cortex‑X925 + 10× Cortex‑A725
CUDA Cores NVIDIA Blackwell Generation
Tensor Cores 5th Generation
RT Cores 4th Generation
Tensor Performance Up to 1 PFLOP
System Memory 128 GB LPDDR5x, coherent unified system memory
Memory Interface 256‑bit
Memory Bandwidth Up to 273 GB/s
Storage 4 TB NVMe M.2 with self‑encryption
USB 4× USB Type‑C
Ethernet 1× RJ‑45 connector, 10 GbE
NIC ConnectX‑7 NIC @ 200 Gbps
Wi‑Fi WiFi 7
Bluetooth BT 5.4 w/ LE
Audio output HDMI multichannel audio output
Power Supply 240 W
GB10 TDP* 140 W
Display Connectors 1× HDMI 2.1a
NVENC | NVDEC 1× | 1×
OS NVIDIA DGX OS
System Dimensions 150 mm L × 150 mm W × 50.5 mm H
System Weight 1.2 kg

*GB10 TDP — declared TDP value for GB10.

Contact an Elmark specialist

Have questions? Need advice? Call or write to us!