Edge AI Solutions

Marbella AI-N

Edge AI Analytics Product powered by NVIDIA L4

Next-Generation Edge Analytics: Silicom Edge AI Marbella Platform and NVIDIA L4 GPU for low-latency edge services

In the data-driven world, the convergence of edge computing and advanced AI capabilities offers efficient and intelligent services. The fusion between the Silicom Marbella platform and the NVIDIA L4 GPU provides unprecedented low-latency edge analytics performance.

Silicom Marbella Platform: The Edge Computing Solution

At the forefront of modern edge computing solutions stands the Silicom Marbella platform, a versatile and resilient networking appliance. Driven by high-performance Intel Xeon-D processors and offering various connectivity options, the Marbella platform creates a scalable environment tailored to the demands of edge analytics. Its compact footprint and robust design make it an ideal fit for diverse edge environments, spanning industries from manufacturing to the Internet of Things (IoT).

NVIDIA L4 GPU: Pioneering AI at the Edge

Complementing the Marbella platform is the NVIDIA L4 GPU, an accelerator designed to redefine edge AI services. Crafted by NVIDIA, a leader in GPU technology, the L4 GPU boasts immense AI processing power, capable of handling intricate neural network algorithms with remarkable speed and efficiency.

Silicom Marbella Visual Analytics EdgeAI nvidia l4

A Synergy that Transforms Edge Analytics

The seamless integration of the Silicom Marbella platform and the NVIDIA L4 GPU culminates in a synergy that propels the edge analytics landscape to new heights. The Marbella platform’s Intel Xeon-D processors lay the foundation for robust data processing and initial analysis. When combined with the NVIDIA L4 GPU, organizations unlock the capacity to execute complex AI tasks at the edge, spanning image recognition, natural language processing, and predictive maintenance.

Technical Excellence, Professional Distinction

• The technical precision offered by the Silicom Marbella platform, and the NVIDIA L4 GPU revolutionizes the landscape of edge analytics across industries:

• Real-Time Insights: Organizations can harness real-time analytics, enabling swift decision-making and agility in response to dynamic market conditions.

• Efficient Resource Utilization: Processing data at the edge minimizes the volume of data transmitted to centralized servers, leading to optimized resource utilization and cost efficiency.

• Operational Efficiency: The Marbella platform and the NVIDIA L4 GPU enhance operational efficiency by reducing latency and facilitating timely insights—critical advantages in sectors such as telecom, manufacturing, healthcare, and transportation.

Edge Analytics: A Professional Edge

In an era shaped by data-driven innovation, the collaboration between the Silicom Marbella platform and the NVIDIA L4 GPU harnesses cutting-edge edge analytics capabilities, empowered by the Marbella platform’s resilient architecture and the NVIDIA L4 GPU’s AI processing capabilities. This partnership transcends technological horizons, enabling organizations to seize real-time insights, optimize operations, and achieve professional excellence in an era defined by data.

Marbella AI-N

Edge AI Analytics Product powered by NVIDIA L4

  • Intel® Xeon® D-1700 Platform. Support 4,8, and 10-core CPU SKUs
  • Memory support up to 256GB DDR4
  • Supports a Dual x4 PCIe NVMe
  • Networking
    • 4x 10G/25G SFP28 (support for 10G RJ45)
    • 4x 1G/10G SFP+
    • 3x 2.5G RJ45
    • 1x 1GbE MGMT
  • Commercial and Industrial Temperature Options
  • Redundant AC or DC Power Supplies
  • Short Depth Enclosure (12”/305mm)
  • PCIe Expansion
  • x16 Gen4 Full-Height ¾-Length
  • x8 Gen3 Half-Height Half-Lenth
  • Support LTE/5G and Timesync modules
  • BMC with in-band support
  • Powered by Nvidia L4 GPU optimized for media stream density and quality.

Marbella AI-N

Edge AI Analytics Product powered by NVIDIA L4

General Technical Specifications

CPU: Intel® Xeon® D-1700. Support for 4,8, and 10-core SKUs
Memory: 3x Channels DDR4, Supports up to 256GB
  • eMMC (4GB to 256GB)
  • Two NVMe (x4 PCIe)
  • Two 2.5” SATA
  • 4x 25G/10G SFP28 (SKU and Configuration Dependent)
  • 4x 10G/1G SFP+ (SKU and Configuration Dependent)
  • 3x 2.5GbE (Intel i226)
  • 1x 1GbE (BMC Management)
AI accelerator:

NVIDIA L4 Tensor Core GPU

  • FP32 – 30.3 teraFLOPs
  • TF32 Tensor Core – 120 teraFLOPS*
  • FP16 Tensor Core – 242 teraFLOPS*
  • BFLOAT16 Tensor Core – 242 teraFLOPS*
  • FP8 Tensor Core – 485 teraFLOPs*
  • INT8 Tensor Core – 485 TOPs*
  • GPU memory – 24GB
  • GPU memory bandwidth – 300GB/s
  • NVENC | NVDEC | JPEG decoders – 2 / 4 / 4
    * Shown with sparsity. Specifications are one-half lower without sparsity.
  • 2x USB3/2
  • Console: Cisco RS232 RJ45
PCIe Expansion:
  • x16 PCIe Gen4 Full Height, ¾ Length Card
  • x8 PCIe Gen3 Half Height, Half Length Card
Cellular (Optional card)
  • Supports optional wireless modules (4G and 5G options)
  • Antenna access through extender
TimeSync (Optional card)
  • External SMA source inputs are 10MHz, PPS, and GNSS/GPS
  • External SMA outputs are 10MHz and PPS
  • AST2620 (no video, option for extended temperature)
  • NCSI to CPU for optional in-band management
  • Thermal monitor/fan control (up to 7)
  • Remote updates
  • Remote Power Management
  • Virtual USB
  • Serial over LAN
  • TPM2.0
  • Hardware Root of Trust
  • Intrusion switch
  • Power / Reset Buttons (programmable)
  • RGB LED’s
  • UEFI BIOS (Consider open-source solution)
  • Optional Redundant BIOS
Power Supply:
  • Redundant 500W Pluggable Power Supply
  • AC and DC Options
  • Commercial SKU: 0C to 40C
  • Industrial SKU: -20C to 65C
Form Factor:
  • 12 inches (305mm) Depth
  • 19” Rackmount

Marbella AI-N

Edge AI Analytics Product powered by NVIDIA L4




80500-0217-G02-SL03A Marbella 10C 64G, NVMe Nvidia L4, 2xAC USA  


YouTube Channel