Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 

Repository files navigation

AI Container Stack

Introduction

This repository provides Dockerfiles and build scripts for creating container images with Intel® Graphics Compute Runtime for oneAPI Level Zero and OpenCL Driver, PyTorch* (with XPU support), and Intel® Open Middleware Xe (Intel® OMIX). The stack is designed for high-performance deep learning workloads on Intel GPUs.

Repository Structure

  • dockerfiles/: contains folders for each image in the stack. Under each folder, you will find one or more Dockerfiles.

About the Stack

The stack consists of the following images:

PyTorch* on Intel® Hardware

dockerfiles/pytorch/

The PyTorch* container images are optimized for Intel® CPUs and GPUs, providing a seamless experience for developers to build and deploy AI applications.

Below is an example command that can be used to build this image:

docker build -t intel/pytorch:xpu-2.11.0-ubuntu24.04 -f dockerfiles/pytorch/xpu-2.11.0-ubuntu24.04.dockerfile .

vLLM with Intel(R) GPUs support

dockerfiles/vllm/

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. This configuration provides vLLM with Intel(R) GPUs support for Intel® Client GPUs and Intel® Data Center GPUs.

Below is an example command that can be used to build this image:

docker build -t intel/vllm:0.21.0-ubuntu24.04 -f dockerfiles/vllm/0.21.0-ubuntu24.04.dockerfile .

Intel® Graphics Compute Runtime for oneAPI Level Zero and OpenCL Driver

dockerfiles/compute-runtime/

The Intel® Graphics Compute Runtime for oneAPI Level Zero and OpenCL Driver is an open source project providing compute API support (Level Zero, OpenCL) for Intel graphics hardware architectures (HD Graphics, Xe).

Below is an example command that can be used to build this image:

docker build -t intel/compute-runtime:26.22.38646.6-devel-ubuntu24.04 -f dockerfiles/compute-runtime/26.22.38646.6-devel-ubuntu24.04.dockerfile .

Intel® Open Middleware Xe

dockerfiles/omix/

Intel® Open Middleware Xe is a set of highly optimized deep learning frameworks and tools for Intel GPUs to accelerate AI workloads.

Below is an example command that can be used to build this image:

docker build -t intel/omix:0.3.0-devel-ubuntu24.04 -f dockerfiles/omix/0.3.0-devel-ubuntu24.04.dockerfile .

Prerequisites

To utilize containers with Intel GPU driver support, the host system must meet the following requirements:

  • Operating System: A supported Linux distribution is required. Refer to Supported Linux Kernels for specific version details.
  • Kernel Mode Driver (KMD): The system must have the appropriate KMD driver installed.

Requirements for AI Containers

If you plan to use the following AI containers, the host system requires the latest kernel version to support the necessary drivers and specific AI features:

  • omix
  • pytorch

Installing the Intel GPU Kernel (Ubuntu)

For optimal compatibility with the latest Intel GPU devices and features, install the kernel from the intel-graphics PPA. Follow the steps below for Ubuntu:

  1. Update package index and install repository management tools:

    sudo apt-get update
    sudo apt-get install -y software-properties-common
  2. Add the Intel Graphics PPA:

    sudo add-apt-repository -y ppa:kobuk-team/intel-graphics
  3. Install the Intel GPU kernel package:

    sudo apt-get update
    sudo apt-get install -y linux-intel

How to provide feedback

Use GitHub Issues for feature requests, bug reports, and minor inquiries. For broader questions and development-related discussions, use GitHub Discussions.

Security

To report a vulnerability, refer to Intel vulnerability reporting policy.

Default container user

In most use-cases we expect that customers may want to install additional software on top of our containers. With this, the default user in these containers would be root. If your flow does not require root (e.g. git clone, build and test your application, push to your project artifactory), we recommend creating a dedicated non-root user, setting up appropriate environment, and running the steps from that user rather than as root.

License

This project is licensed under the terms of the MIT license. See LICENSE for more information.

Notices and Disclaimers

© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. OpenCL and the OpenCL logo are trademarks of Apple Inc. used by permission by Khronos. Other names and brands may be claimed as the property of others.

About

No description, website, or topics provided.

Resources

Code of conduct

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Used by

Contributors

Languages