Intel XPU Software Installation

This guide shows how to use an Intel® Extension for TensorFlow* XPU package, which provides GPU and CPU support simultaneously.

Hardware Requirements

Verified Hardware Platforms:

Intel® Data Center GPU Max Series, Driver Version: 736
Intel® Data Center GPU Flex Series 170, Driver Version: 736
Experimental: Intel® Arc™ A-Series

For experimental support of the Intel® Arc™ A-Series GPUs, please refer to Intel® Arc™ A-Series GPU Software Installation for details.

Software Requirements

Ubuntu 22.04, Red Hat 8.6 (64-bit)
- Intel® Data Center GPU Flex Series
Ubuntu 22.04, Red Hat 8.6 (64-bit), SUSE Linux Enterprise Server(SLES) 15 SP4/SP5
- Intel® Data Center GPU Max Series
Intel® oneAPI Base Toolkit 2024.0
TensorFlow 2.14.0
Python 3.9-3.11
pip 19.0 or later (requires manylinux2014 support)

Install GPU Drivers

OS	Intel GPU	Install Intel GPU Driver
Ubuntu 22.04, Red Hat 8.6	Intel® Data Center GPU Flex Series	Refer to the Installation Guides for latest driver installation. If install the verified Intel® Data Center GPU Max Series/Intel® Data Center GPU Flex Series 736, please append the specific version after components, such as `sudo apt-get install intel-opencl-icd==23.30.26918.50-736~22.04`
Ubuntu 22.04, Red Hat 8.6, SLES 15 SP4/SP5	Intel® Data Center GPU Max Series	Refer to the Installation Guides for latest driver installation. If install the verified Intel® Data Center GPU Max Series/Intel® Data Center GPU Flex Series 736, please append the specific version after components, such as `sudo apt-get install intel-opencl-icd==23.30.26918.50-736~22.04`

Install via Docker container

The Docker container includes the Intel® oneAPI Base Toolkit, and all other software stack except Intel GPU Drivers. Install the GPU driver in host machine bare metal environment, and then launch the docker container directly.

Build Docker container from Dockerfile

Run the following Dockerfile build procedure to build the pip based deployment container.

Get docker container from dockerhub

Pre-built docker images are available at DockerHub. Run the following command to pull Intel® Extension for TensorFlow* Docker container image (xpu) to your local machine.

$ docker pull intel/intel-extension-for-tensorflow:xpu
$ docker run -it -p 8888:8888 --device /dev/dri -v /dev/dri/by-path:/dev/dri/by-path intel/intel-extension-for-tensorflow:xpu

To use Intel® Optimization for Horovod* with the Intel® oneAPI Collective Communications Library (oneCCL), pull Intel® Extension for TensorFlow* Docker container image (xpu) to your local machine and set the required environment variables after creating the container by the following command.

$ docker pull intel/intel-extension-for-tensorflow:xpu
$ docker run -it -p 8888:8888 --device /dev/dri -v /dev/dri/by-path:/dev/dri/by-path --ipc=host intel/intel-extension-for-tensorflow:xpu
$ export LD_LIBRARY_PATH=/opt/intel/oneapi/redist/opt/mpi/libfabric/lib:$LD_LIBRARY_PATH
$ export PATH=/opt/intel/oneapi/redist/bin:$PATH
$ export I_MPI_ROOT=/opt/intel/oneapi/redist/lib
$ export CCL_ROOT=/opt/intel/oneapi/redist
$ export FI_PROVIDER_PATH=/opt/intel/oneapi/redist/opt/mpi/libfabric/lib/prov

Then go to your browser on http://localhost:8888/

Install via PyPI wheel in bare metal

Install oneAPI Base Toolkit Packages

Need to install components of Intel® oneAPI Base Toolkit:

Intel® oneAPI DPC++ Compiler
Intel® oneAPI Math Kernel Library (oneMKL)
Intel® oneAPI Threading Building Blocks (TBB), dependency of DPC++ Compiler.
Intel® oneAPI Collective Communications Library (oneCCL), required by Intel® Optimization for Horovod* only

$ wget https://registrationcenter-download.intel.com/akdlm//IRC_NAS/20f4e6a1-6b0b-4752-b8c1-e5eacba10e01/l_BaseKit_p_2024.0.0.49564.sh
# 3 components are necessary: DPC++/C++ Compiler, DPC++ Libiary and oneMKL
# if you want to run distributed training with Intel® Optimization for Horovod*, oneCCL is needed too(Intel® oneAPI MPI Library will be installed automatically as its dependency)
$ sudo sh l_BaseKit_p_2024.0.0.49564.sh

For any more details, follow the procedure in https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit.html.

Setup environment variables

# DPC++ Compiler/oneMKL
source /path to basekit/intel/oneapi/compiler/latest/env/vars.sh
source /path to basekit/intel/oneapi/mkl/latest/env/vars.sh

# oneCCL (and Intel® oneAPI MPI Library as its dependency), required by Intel® Optimization for Horovod* only
source /path to basekit/intel/oneapi/mpi/latest/env/vars.sh
source /path to basekit/intel/oneapi/ccl/latest/env/vars.sh

You may install more components than Intel® Extension for TensorFlow* needs, and if required, setvars.sh can be customized to point to a specific directory by using a configuration file:

source /opt/intel/oneapi/setvars.sh --config="full/path/to/your/config.txt"

Install TensorFlow

The Python development and virtual environment setup recommendation by TensorFlow to isolate package installation from the system.

The Intel® Extension for TensorFlow* requires stock TensorFlow, and the version should be == 2.14.0.

Virtual environment install

You can follow the instructions in stock tensorflow install to activate the virtual environment.

On Linux, it is often necessary to first update pip to a version that supports manylinux2014 wheels.

(tf)$ pip install --upgrade pip

To install in virtual environment, you can run

(tf)$ pip install tensorflow==2.14.0

System environment install

If you prefer install tensorflow in $HOME, please append --user to the commands.

$ pip install --user tensorflow==2.14.0

And the following system environment install for Intel® Extension for TensorFlow* will also append --user to the command.

Install Intel® Extension for TensorFlow*

To install a XPU version in virtual environment, which depends on Intel GPU drivers and oneAPI BaseKit, you can run

(tf)$ pip install --upgrade intel-extension-for-tensorflow[xpu]

Check the Environment for XPU

(tf)$ bash /path to site-packages/intel_extension_for_tensorflow/tools/env_check.sh

Verify the Installation

python -c "import intel_extension_for_tensorflow as itex; print(itex.__version__)"

Then, you can get the information that both CPU and GPU backends are loaded successfully from the console log.

2023-07-28 12:00:00.374832: I itex/core/wrapper/itex_cpu_wrapper.cc:42] Intel Extension for Tensorflow* AVX512 CPU backend is loaded.
2023-07-28 12:00:00.217981: I itex/core/wrapper/itex_gpu_wrapper.cc:35] Intel Extension for Tensorflow* GPU backend is loaded.

NOTE: If Intel® Extension for TensorFlow* XPU package is installed on GPU support platform, both CPU and GPU backends will be loaded as pluggable device via TensorFlow. GPU backend will be activated as default backend.

XPU for CPU only platform

If Intel® Extension for TensorFlow* XPU package is installed on CPU only platform, only CPU backend will be loaded. Please refer to Intel CPU Software Installation Hardware Requirements and Software Requirements for the platform requirements.

Verify the Installation

python -c "import intel_extension_for_tensorflow as itex; print(itex.__version__)"

Then, you can get the information that only CPU backend is loaded successfully from the console log.

2023-07-28 12:00:00.205706: I itex/core/wrapper/itex_cpu_wrapper.cc:42] Intel Extension for Tensorflow* AVX512 CPU backend is loaded.
2023-07-28 12:00:00.313231: E itex/core/wrapper/itex_gpu_wrapper.cc:49] Could not load Intel Extension for Tensorflow* GPU backend, GPU will not be used.
If you need help, create an issue at https://github.com/intel/intel-extension-for-tensorflow/issues