intel/pytorch repository overview

⁠PyTorch* on Intel® Hardware

⁠Intel® CPUs support for PyTorch*

On Intel® CPUs, PyTorch* takes advantage of the following instruction sets:

• Intel® Advanced Matrix Extensions (Intel® AMX)
• Intel® Advanced Vector Extensions 512 (Intel® AVX-512)
• Vector Neural Network Instructions (VNNI)

Images available here start with the Ubuntu* 24.04⁠ base image with PyTorch*⁠ for Intel® CPUs.

docker run -it --rm \
    --ipc=host \
    intel/pytorch:cpu-2.13.0

⁠Intel® GPUs support for PyTorch*

The following Intel® GPUs are supported by PyTorch* through the xpu device backend:

• Intel® Arc A-Series Graphics
• Intel® Arc B-Series Graphics
• Intel® Core™ Ultra Processors with Intel® Arc™ Graphics
• Intel® Core™ Ultra Processors (Series 2) with Intel® Arc™ Graphics
• Intel® Core™ Ultra Mobile Processors (Series 2) with Intel® Arc™ Graphics
• Intel® Core™ Ultra Mobile Processors (Series 3) with Intel® Arc™ Graphics

Above GPU platforms require different minimal versions of Linux kernel. For detailed requirements see https://dgpu-docs.intel.com⁠. For usage examples running PyTorch* with xpu device backend see https://docs.pytorch.org/docs/stable/notes/get_start_xpu.html⁠.

Images available here start with the Intel® Graphics Compute Runtime⁠ base image, which is itself based on Ubuntu* 24.04⁠, with PyTorch*⁠ for Intel® GPUs.

DEVICE=${DEVICE:-/dev/dri/renderD128}
DEVICE_GRP=$(stat --format %g $DEVICE)
docker run -it --rm \
    --device $DEVICE \
    --group-add $DEVICE_GRP \
    -v /dev/dri/by-path:/dev/dri/by-path \
    --ipc=host \
    intel/pytorch:latest

⁠Known Limitations

The following Intel GPUs are not fully validated with the provided container images and are not guaranteed to work:

• Intel® Data Center GPU Max Series
• Intel® Data Center GPU Flex Series

⁠Notes on the bundled packages

The images bundle torch, torchvision and torchaudio installed from the official PyTorch* wheel index (https://download.pytorch.org/whl/cpu and https://download.pytorch.org/whl/xpu):

torchaudio has no release matching PyTorch* 2.13.0; 2.11.0 is the latest published wheel.

⁠Dockerfiles

The Dockerfiles used to build the container images are available in intel/containers⁠ repository.

⁠License Agreement

By downloading and using these container images and the included software, you agree to the terms and conditions of the software license agreements⁠.

In accordance with the terms and conditions of these licenses, particularly those requiring source code availability (such as GPL), the source code for all such open-source components included in these container images can be obtained from here⁠.

⁠Notes

All Ubuntu container images are intended for demo purposes only and not intended for production use. To receive expanded security maintenance from Canonical on the Ubuntu base layer, you may follow the how-to guide to enable Ubuntu Pro in a Dockerfile⁠ which will require the image to be rebuilt.

⁠Acknowledgements

Intel and the Intel logo 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.