Getting Started

Installation

Numba-dpex depends on following components:

It is recommended to use conda packages from Intel Distribution for Python channel or anaconda.org/intel channel. Intel Distribution for Python is available from Intel oneAPI.

Create conda environment:

export ONEAPI_ROOT=/opt/intel/oneapi
conda create -n numba-dpex-env numba-dpex dpnp -c ${ONEAPI_ROOT}/conda_channel

Build and Install Conda Package

Create and activate conda build environment:

conda create -n build-env conda-build
conda activate build-env

Set environment variable ONEAPI_ROOT and build conda package:

export ONEAPI_ROOT=/opt/intel/oneapi
conda build conda-recipe -c ${ONEAPI_ROOT}/conda_channel

Install conda package:

conda install numba-dpex

Build and Install with setuptools

setup.py requires environment variable ONEAPI_ROOT and following packages installed in conda environment:

export ONEAPI_ROOT=/opt/intel/oneapi
conda create -n numba-dpex-env -c ${ONEAPI_ROOT}/conda_channel python=3.7 dpctl dpnp numba spirv-tools dpcpp-llvm-spirv llvmdev pytest
conda activate numba-dpex-env

Activate DPC++ compiler:

source ${ONEAPI_ROOT}/compiler/latest/env/vars.sh

For installing:

python setup.py install

For development:

python setup.py develop

Build and Install with docker

docker run --rm -it \
-v /path/to/numba-dpex/source:/build \
-v /path/to/dist:/dist
ghcr.io/intelpython/numba-dpex/builder:0.20.0-py3.10

python setup.py develop
python setup.py bdist_wheel
cp dist/numba_dpex*.whl /dist/

Now you can install numba-dpex wheel in whatever compatible environment with pip. You will find numba_dpex*.whl file in the /path/to/dist location in your host system.

You can check what dpctl and dpnp is shipped with builder by running pip list. In case you need another version, consider building builder target with necessary build args. Refer to Docker section for more details.

Testing

See folder numba_dpex/tests.

To run the tests:

python -m pytest --pyargs numba_dpex.tests

Examples

See folder numba_dpex/examples.

To run the examples:

python numba_dpex/examples/sum.py

Docker

An easy way you can try numba_dpex is by using docker. To try out numba dpex simply run:

docker run --rm -it ghcr.io/intelpython/numba-dpex/runtime:0.20.0-py3.10

import dpctl

dpctl.lsplatform()

Refer to Docker section for more options.