new

XpertStation WS300 NVIDIA DGX Station

XpertStation WS300 on NVIDIA DGX Station Architecture

Manufacturer: MSI EPS
Manufacturer: MSI EPS
  • Desktop system class AI supercomputer (DGX-class)

  • Grace Blackwell Superchip (GB300) - integrated CPU + GPU

  • ~784 GB of unified memory - working with the largest models

  • Performance enabling LLLM training and inference locally

  • Support for advanced workloads: AI, simulations, digital twins

  • NVLink-C2C - direct CPU-GPU communication

  • ConnectX-8 (up to 800 Gbps) - scaling and high-speed communications

  • Platform for AI teams, data science and R&D

Product out of stock

Free shipping from €300

Promocja cenowa na model HDR-15-5

Product intended for professional use only
XpertStation WS300 NVIDIA DGX Station

XpertStation WS300 NVIDIA DGX Station

Description

 
AI Supercomputer Workstation

XpertStation WS300

Data centre class power brought to the desktop. A system designed for the most demanding AI, simulation and computing environments.

New generation desktop AI

XpertStation WS300 is a system based on the NVIDIA DGX Station architecture, delivering data center-class performance directly to the workstation.

With CPU and GPU integration and massive memory, the platform eliminates the limitations of traditional workstations.

XpertStation WS300
~784 GB

unified memory

GB300

Grace Blackwell Superchip

800 Gbps

networking ConnectX-8

Grace Blackwell Ultra

Grace Blackwell Ultra

The system uses a GB300 Grace Blackwell Ultra Superchip that connects the CPU and GPU via NVLink-C2C.

This provides direct access to shared memory and significantly higher bandwidth than traditional architectures.

Memory for large AI models

XpertStation WS300 offers a huge memory footprint, combining HBM3e and LPDDR5X in one cohesive architecture.

HBM3e
for GPU
LPDDR5X
for CPU
Unified memory
no data copying

Data centre class performance

The platform enables local training and running of AI models with performance comparable to HPC infrastructure.

  • LLM and generative AI
  • simulation and digital twins
  • data science
  • autonomous systems
AI performance

Scaling and communication

NVIDIA ConnectX-8 provides up to 800 Gbps of bandwidth, enabling systems to be connected and larger AI environments to be built.

AI without limits

XpertStation WS300 brings the power of data center infrastructure directly to the workstation.

.

Technical Specification

CPU

NVIDIA Grace CPU Superchip with 72 Arm Neoverse V2 cores

GPU

Single NVIDIA Blackwell Ultra GPU

Memory

Up to 775 GB of Coherent Memory

- CPU Memory Up to 496GB LPDDR5X | Up to 396 GB/s

- GPU Memory Up to 279GB HBM3e | 8 TB/s

Networking

(2) QSFP112 ports with NVIDIA ConnectX-8 SuperNICs

Power supply

(1) 1600W ATX 80 PLUS Titanium

 

 

Contact an Elmark specialist

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