SOLD OUT
nvidia dgx spark gpu powered ai workstation main
nvidia dgx spark gpu powered ai workstation ports
nvidia dgx spark gpu powered ai workstation side
nvidia dgx spark gpu powered ai workstation top
nvidia dgx spark gpu powered ai workstation connection
/ AI & Work Station HPC / Nvidia DGX / NVIDIA DGX Spark GPU-Powered AI Workstation

NVIDIA DGX Spark GPU-Powered AI Workstation

650,000
581,000

Key Features

  • NVIDIA DGX Spark
  • Blackwell GPU
  • 128 GB LPDDR5x
  • 4 TB NVMe Storage
  • 200 Gbps NIC
  • HDMI 2.1a Output

Out Of Stock

Share:

Technical Specifications

Architecture & CPU

ArchitectureNVIDIA Grace BlackwellCPU20-core Arm (10x Cortex-X925 + 10x Cortex-A725)GPUBlackwell ArchitectureCUDA CoresBlackwell GenerationTensor Cores5th GenerationRT Cores4th Generation

Performance & Memory

Tensor PerformanceUp to 1 PFLOP FP4System Memory128 GB LPDDR5x, coherent unified memoryMemory Interface256-bitMemory Bandwidth273 GB/s

Storage & Expansion

Storage4 TB NVMe M.2, self-encryptionUSB Ports4x USB Type-CNVENC / NVDEC1x / 1x

Networking & Connectivity

Ethernet1x RJ-45, 10 GbENICConnectX-7 @ 200 GbpsWi-FiWiFi 7BluetoothBT 5.4

Display & Audio

Display Connector1x HDMI 2.1aAudioHDMI multichannel audio output

Power & Dimensions

Power Supply240 WGB10 TDP140 WSystem Dimensions150 x 150 x 50.5System Weight1.2 kg

Operating System

Supported SystemNVIDIA DGX™ OS

Warranty Information

WarrantyAs Per PCB Store Policy

Product Description

NVIDIA DGX Spark GPU-Powered AI Workstation

If your AI workloads are slowing down due to outdated hardware, the NVIDIA DGX Spark GPU-Powered AI Workstation delivers unmatched speed and efficiency, engineered to accelerate even the most demanding AI models. Designed for researchers, engineers, and AI developers, it harnesses the power of the Blackwell GPU architecture to manage large-scale deep learning, data analysis, and high-performance computing tasks with precision.

Its tightly integrated CPU and GPU, combined with 128 GB of unified LPDDR5x memory and ultra-fast NVMe storage, ensures massive datasets are processed rapidly and reliably. This eliminates bottlenecks and delays, allowing complex AI workflows to run smoothly and efficiently.

This workstation belongs to the Desktop PC category, offering the power and reliability expected from a professional-grade desktop system. It is also classified under the Portable Mini PC sub-category, providing compact form factor options without compromising performance. Additionally, it falls under the Workstation PC sub-category, ideal for developers and AI researchers handling heavy computational tasks.

Beyond traditional computing classifications, it is part of the AI & Workstation HPC category, designed for high-performance AI research, deep learning, and scientific computing. Finally, within NVIDIA’s ecosystem, it belongs to the specialized NVIDIA DGX sub-category, making it a premier choice for professional machine learning and AI deployments in Bangladesh.

NVIDIA DGX Spark Workstation Specifications and Performance Overview

The DGX Spark delivers a high-performance computing environment for AI model training and inference. Its 20-core Arm CPU, combining 10 Cortex-X925 and 10 Cortex-A725 cores, works seamlessly with the Blackwell GPU architecture. With 128 GB of unified LPDDR5x system memory and 4 TB NVMe storage, it manages complex AI workloads without bottlenecks.

As a Desktop PC and Workstation PC, it integrates multiple connectivity options, including WiFi 7, Bluetooth 5.4, and a 10 GbE NIC. This makes it ideal for multi-user setups, cloud integration, or localized AI research, all within a compact 1.2 kg frame.

Key Features and Advantages of NVIDIA DGX Spark

  • Blackwell GPU Architecture powers AI with unmatched tensor and RT core efficiency.
  • Unified System Memory ensures faster data transfers between CPU and GPU.
  • High-Capacity Storage supports large datasets with NVMe M.2 self-encryption.
  • Ultra-Fast Networking with 200 Gbps ConnectX-7 NIC reduces latency in distributed AI workflows.
  • Compact Desktop Form Factor fits anywhere while maintaining professional workstation standards.

Leveraging NVIDIA’s expertise in GPU acceleration, the DGX Spark brings real-world AI performance, enabling data scientists and developers to run massive deep learning models directly on a desktop workstation.

Expand Your AI Workstation with Compatible Components

The NVIDIA DGX Spark GPU-Powered AI Workstation doesn’t just deliver raw AI power—it works best when paired with the right components. Its Blackwell GPU cores handle massive calculations, while high-speed NVMe SSDs store and access large datasets instantly. The 128 GB LPDDR5x memory works in tandem with PCIe expansion cards to manage heavy AI models efficiently. Networking is streamlined with the ConnectX-7 NIC and 10 GbE Ethernet, making multi-GPU or distributed setups smooth and reliable. For display and visualization, the workstation supports HDMI 2.1a outputs, which can be paired with professional monitors and audio systems for precise monitoring. Even smaller components like Type-C USB hubs and cooling modules play a role in maintaining performance stability. Together, these related components ensure the DGX Spark can grow with your AI projects, giving developers and researchers a flexible, high-performance ecosystem.

Power, Efficiency, and Technical Highlights

The DGX Spark operates on a 240 W power supply with a 140 W GB10 TDP. Optimized for low thermal output, it maintains peak performance without overheating, even under continuous high-load AI computations. Its coherent 128 GB system memory and 273 GB/s bandwidth deliver sustained throughput for large neural networks.

The combination of 5th-generation Tensor Cores and 4th-generation RT cores ensures maximum performance in AI training, inference, and simulation tasks. FP4 tensor operations reach up to 1 PFLOP, providing computational power previously limited to server-grade systems.

Compatibility and Use Cases

The NVIDIA DGX Spark is compatible with a broad range of AI frameworks and GPU-accelerated software. Its PCIe Gen 5 architecture supports NVIDIA NVENC/NVDEC for video encoding, while the HDMI 2.1a port allows multichannel display setups. Developers can integrate it with cloud platforms, other Workstation PCs, or multiple DGX Spark units for distributed learning.

Ideal for AI research, scientific simulations, autonomous vehicle model training, and natural language processing workloads, it performs seamlessly alongside GPUs like the NVIDIA RTX and A100 series when deployed in hybrid workstation environments.

Reliability and Warranty

The DGX Spark is backed by NVIDIA DGX OS and comes with robust reliability features. Its self-encrypting storage safeguards critical datasets. With compact form factors and engineered thermal efficiency, this Desktop PC ensures consistent uptime and reduced maintenance.

Warranty services in Bangladesh are supported through PCB Store, giving customers secure access to repairs, updates, and technical assistance for all workstation components.

Where to Buy the NVIDIA DGX Spark in Bangladesh

PCB Store in Bangladesh offers the NVIDIA DGX Spark GPU-Powered AI Workstation for AI professionals and developers seeking top-tier computing power. Limited availability makes it essential to secure your workstation promptly. Combine high-performance GPUs, unified system memory, and low-latency networking to elevate AI development.

Visit PCB Store to purchase your DGX Spark today and bring professional AI computation to your Desktop PC setup in Bangladesh.

Delivery Charge Info

Inside Dhaka: 350

Outside Dhaka: 400

FAQs

  • Is the DGX Spark compatible with Ubuntu or Windows OS?

    It runs NVIDIA DGX OS natively but supports containerized AI frameworks compatible with Linux distributions like Ubuntu. Windows is not officially supported.

  • Can I upgrade the memory on the DGX Spark?

    No, the 128 GB LPDDR5x memory is soldered and unified for optimized GPU-CPU performance.

  • What is the thermal output under full AI load?

    The workstation maintains low thermal output, with a GB10 TDP of 140 W, suitable for continuous AI workloads.

  • Does it support multiple displays?

    Yes, HDMI 2.1a allows multichannel output for complex visualization and research environments.

  • How fast is the networking on this workstation?

    It features a 200 Gbps ConnectX-7 NIC and 10 GbE Ethernet for ultra-low latency connections.

  • DGX Spark vs NVIDIA RTX workstation: which is faster?

    DGX Spark exceeds standard RTX workstations in FP4 tensor operations, delivering up to 1 PFLOP performance.

  • How much power does it consume?

    The system uses 240 W, with 140 W GB10 TDP, balancing energy efficiency and performance.

  • What is the price and availability at PCB Store Bangladesh?

    The NVIDIA DGX Spark is available at PCB Store, Bangladesh, with current pricing provided directly through their sales team.

  • Can it handle large AI datasets efficiently?

    Yes, 4 TB NVMe storage and 128 GB unified memory manage extensive AI datasets with minimal latency.

Popular Products