Overview
CUDA®-enabled x86 Container
About Advantech Container Catalog
Advantech Container Catalog is a comprehensive collection of ready-to-use, containerized software packages designed to accelerate the development and deployment of Edge AI applications. By offering pre-integrated solutions optimized for embedded hardware, it simplifies the challenges often faced with software and hardware compatibility, especially in GPU/NPU-accelerated environments.
Key benefits of the Container Catalog include:
| Feature / Benefit | Description |
|---|---|
| Accelerated Edge AI Development | Ready-to-use containerized solutions for fast prototyping and deployment |
| Hardware Compatibility Solved | Eliminates embedded hardware and AI software package incompatibility |
| GPU/NPU Access Ready | Supports passthrough for efficient hardware acceleration |
| Model Conversion & Optimization | Built-in AI model quantization and format conversion support |
| Optimized for CV & LLM Applications | Pre-optimized containers for computer vision and large language models |
| Open Ecosystem | 3rd-party developers can integrate new apps to expand the platform |
Container Overview
CUDA®-enabled x86 Container is a base image that provides a pre-configured NVIDIA CUDA® Toolkit development environment with NVIDIA GPU passthrough capabilities, designed for industrial AI and machine learning applications. This solution simplifies the deployment of GPU-accelerated workloads in edge computing environments while ensuring consistent, reproducible results across development and production environments.
Key Features
Optimized for AI development on NVIDIA GPU, this container provides flexible NVIDIA CUDA® Toolkit/cuDNN configurations, full GPU passthrough, and essential tools pre-installed for rapid deployment.
| Feature | Content |
|---|---|
| Automatic CUDA® Detection | Automatically detects installed NVIDIA CUDA® Toolkit version from your system |
| Flexible Configuration | Supports command-line configuration of NVIDIA CUDA® Toolkit and cuDNN versions |
| Multi-Version Support | Compatible with NVIDIA CUDA® Toolkit versions from 11.8 to 12.4 and corresponding cuDNN libraries |
| Complete GPU Passthrough | Utilizes all available NVIDIA GPUs with full compute capabilities |
| Optimized Memory Management | Configured with 8GB shared memory for high-performance GPU operations |
Application Support
The Advantech CUDA®-enabled x86 Container container is tailored for high-performance AI and ML workloads. It supports a wide range of frameworks and libraries optimized for GPU acceleration in edge and industrial environments:
| Application | Content |
|---|---|
| Deep Learning Frameworks | TensorFlow, PyTorch, JAX, MXNet |
| Computer Vision Libraries | OpenCV, NVIDIA TensorRT |
| HPC Applications | CUDA®-accelerated scientific computing |
| Data Science | NumPy, SciPy, scikit-learn with GPU acceleration |
| NLP Processing | BERT, GPT, and transformer models |
| Industrial Applications | Edge AI, Computer Vision, Predictive Maintenance |
Host Device Prerequisites
Users should ensure that the selected Advantech host device meets the specifications below and has the required software packages installed before running this container image.
| Item | Specification |
|---|---|
| Compatible Hardware | Advantech x86 Edge Devices (tested on AIR-520) - refer to Compatible hardware |
| NVIDIA GPU | Host system with NVIDIA GPU(s) (tested on RTX 6000 Ada Generation) |
| NVIDIA driver | 550.120 or later, compatible with NVIDIA CUDA® Toolkit 11.8+ |
| Docker & Docker Compose | both versions > 2.0 |
| Recommended Device OS | Linux-based operating system (Ubuntu 20.04+) |
| NVIDIA Container Toolkit | Installing the NVIDIA Container Toolkit |
| SUSI IoT Installation | SUSI drivers installed on host system SUSI IoT Installation Guide |
Validated Hardware Platform
The CUDA®-enabled x86 Container has been thoroughly tested and validated on Advantech AIR-520 with detailed hardware specification below:
| Item | Specification |
|---|---|
| CPU | AMD EPYC 7543P 32-Core Processor (64 logical cores) |
| Memory | 251GB RAM (246GB available) |
| GPU | 2x NVIDIA RTX 6000 Ada Generation (48GB VRAM each) |
| Operating System | Ubuntu 22.04.5 LTS (Jammy Jellyfish) |
| Kernel Version | 6.8.0-59-generic |
| NVIDIA Driver | 550.144.03 |
| NVIDIA CUDA® Toolkit Version | 12.2 |
| cuDNN Version | 8.9.7 |
Container Environment Overview
Pre-installed Development Tools on Container Image
The CUDA®-enabled x86 Container container comes with these essential development tools pre-installed:
| Type | Components |
|---|---|
| Build Tools | gcc/g++, make, cmake, build-essential |
| Version Control | git |
| Python Environmen | Python 3 with pip |
| Utilities | wget, curl, vim, unzip |
| System Tools | ca-certificates, gnupg2, lsb-release, software-properties-common |
GPU Passthrough Capabilities
The Advantech L2-01 container supports full GPU passthrough for:
- NVIDIA RTX/Quadro series (tested with RTX 6000 Ada Generation)
- NVIDIA Tesla/A-series accelerators
- NVIDIA GeForce RTX/GTX series
- Multi-GPU configurations with automatic load balancing
- Support for NVIDIA CUDA® Toolkit compute capability 8.x to 9.x devices
The container can handle:
- Up to 8 GPUs simultaneously
- Up to 96GB VRAM per GPU
- Mixed GPU types in the same system
- Dynamic GPU allocation
Container Quick Start Guide
For container quick start, including docker-compose file, and more, please refer to Advantech EdgeSync Container Repository
Test Results
The diagnostic report confirms full functionality of all GPU components on this container, showcaseing GPU access passthroguh is fully enabled:
====== FINAL REPORT ======
✅ ALL CHECKS PASSED: Your NVIDIA CUDA® Toolkit environment appears to be properly configured
┌─────────────────────────────────────────────────────┐
│ NVIDIA CUDA® Toolkit Environment Summary │
├───────────────────────┬─────────────────────────────┤
│ Overall Status │ PASS │
├───────────────────────┼─────────────────────────────┤
│ NVIDIA Driver │ 550.144.03 │
│ NVIDIA CUDA® Toolkit Version │ 12.2 │
│ cuDNN Version │ 8.9.7.* │
│ GPU Count │ 2 │
├───────────────────────┼─────────────────────────────┤
│ Driver Status │ ✓ Passed │
│ NVIDIA CUDA® Toolkit Toolkit Status │ ✓ Passed │
│ cuDNN Status │ ✓ Passed │
│ NVIDIA CUDA® Toolkit Test Status │ ✓ Passed │
└───────────────────────┴─────────────────────────────┘
GPU Performance Validation
Detail of detected GPUs showing full hardware access and acceleration capabilities:
Device 0: NVIDIA RTX 6000 Ada Generation
Compute capability: 8.9
Total global memory: 47.50 GB
Multiprocessors: 142
Device 1: NVIDIA RTX 6000 Ada Generation
Compute capability: 8.9
Total global memory: 47.50 GB
Multiprocessors: 142
NVIDIA Driver Performance
The container has passed all driver functionality tests with full access to the underlying NVIDIA hardware:
Wed May 7 02:32:35 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.144.03 Driver Version: 550.144.03 NVIDIA CUDA® Toolkit Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA RTX 6000 Ada Gene... Off | 00000000:01:00.0 Off | Off |
| 33% 62C P8 41W / 300W | 12MiB / 49140MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
| 1 NVIDIA RTX 6000 Ada Gene... Off | 00000000:81:00.0 On | Off |
| 34% 62C P8 34W / 300W | 334MiB / 49140MiB | 4% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
Copyright © 2025 Advantech Corporation. All rights reserved.