Skip to content
Linux CUDA x64 Build

Initialize DotNet package validation pipeline #2211

Sign in to view logs

Initialize DotNet package validation pipeline

Initialize DotNet package validation pipeline #2211

Workflow file for this run

.github/workflows/linux-gpu-x64-build.yml at 2820308
name: "Linux CUDA x64 Build"
on:
workflow_dispatch:
push:
branches:
- main
- rel-*
pull_request:
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
cancel-in-progress: true
env:
ORT_NIGHTLY_REST_API: "https://feeds.dev.azure.com/aiinfra/PublicPackages/_apis/packaging/Feeds/ORT-Nightly/packages?packageNameQuery=Microsoft.ML.OnnxRuntime.Gpu.Linux&api-version=6.0-preview.1"
ORT_PACKAGE_NAME: Microsoft.ML.OnnxRuntime.Gpu.Linux
ORT_NIGHTLY_SOURCE: "https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/nuget/v3/index.json"
DOTNET_INSTALL_DIR: "${{ github.workspace }}/dotnet"
jobs:
linux-cuda-x64-build:
env :
PYTHON_EXECUTABLE: "/opt/python/cp38-cp38/bin/python3.8"
runs-on: [ "self-hosted", "1ES.Pool=onnxruntime-genai-Ubuntu2004-T4" ]
steps:
- name: Checkout OnnxRuntime GenAI repo
uses: actions/checkout@v4
with:
submodules: true
# We are using the same manylinux repo as the one used in the packaging build
- name: Checkout ManyLinux repo
uses: actions/checkout@v4
with:
repository: pypa/manylinux
ref: 5eda9aded5462201e6310105728d33016e637ea7
clean: true
path: manylinux
submodules: true
- name: Install jq
run: |
sudo apt-get install jq
- uses: actions/setup-dotnet@v4
with:
dotnet-version: '8.0.x'
- name: Download OnnxRuntime
run: |
ORT_NIGHTLY_VERSION=$(curl -s "${{ env.ORT_NIGHTLY_REST_API }}" | jq -r '.value[0].versions[0].normalizedVersion')
echo "$ORT_NIGHTLY_VERSION"
echo "ORT_NIGHTLY_VERSION=$ORT_NIGHTLY_VERSION" >> $GITHUB_ENV
- name: Download OnnxRuntime Nightly
run: |
dotnet new console
dotnet add package ${{ env.ORT_PACKAGE_NAME }} --version ${{ env.ORT_NIGHTLY_VERSION }} --source ${{ env.ORT_NIGHTLY_SOURCE }} --package-directory .
dotnet build
continue-on-error: true
- name: list files
shell: bash
run: |
ls -l
ls -R ${{ env.ORT_PACKAGE_NAME }}
continue-on-error: true
# TODO: Find out why do we need to to have libonnxruntime.so.$ort_version
- name: Extract OnnxRuntime library and header files
run: |
set -e -x
mkdir -p ort/lib
mv microsoft.ml.onnxruntime.gpu.linux/${{ env.ORT_NIGHTLY_VERSION }}/buildTransitive/native/include ort/
mv microsoft.ml.onnxruntime.gpu.linux/${{ env.ORT_NIGHTLY_VERSION }}/runtimes/linux-x64/native/* ort/lib/
ort_version=$(echo ${{ env.ORT_NIGHTLY_VERSION }} | cut -d- -f1-1)
cp ort/lib/libonnxruntime.so ort/lib/libonnxruntime.so.$ort_version
- name: Get Docker Image
run: |
set -e -x
az login --identity --username 63b63039-6328-442f-954b-5a64d124e5b4
az acr login --name onnxruntimebuildcache --subscription 00c06639-6ee4-454e-8058-8d8b1703bd87
python3 tools/ci_build/get_docker_image.py --dockerfile tools/ci_build/github/linux/docker/manylinux/Dockerfile.manylinux2_28_cuda_11.8 \
--context tools/ci_build/github/linux/docker/manylinux \
--docker-build-args "--build-arg BUILD_UID=$( id -u )" \
--container-registry onnxruntimebuildcache \
--manylinux-src manylinux \
--multiple_repos \
--repository onnxruntimecudabuildx64
- name: Config with Cmake in Docker
run: |
set -e -x
docker run \
--gpus all \
--rm \
--volume $GITHUB_WORKSPACE:/ort_genai_src \
-w /ort_genai_src onnxruntimecudabuildx64 \
bash -c " \
/usr/bin/cmake --preset linux_gcc_cuda_release \
-DMANYLINUX=ON \
-DPYTHON_EXECUTABLE=${{ env.PYTHON_EXECUTABLE }} "
- name: Build with Cmake in Docker
run: |
set -e -x
docker run \
--gpus all \
--rm \
--volume $GITHUB_WORKSPACE:/ort_genai_src \
-w /ort_genai_src onnxruntimecudabuildx64 \
bash -c " \
/usr/bin/cmake --build --preset linux_gcc_cuda_release"
- name: Get HuggingFace Token
run: |
az login --identity --username 63b63039-6328-442f-954b-5a64d124e5b4
HF_TOKEN=$(az keyvault secret show --vault-name anubissvcsecret --name ANUBIS-HUGGINGFACE-TOKEN --query value)
echo "::add-mask::$HF_TOKEN"
echo "HF_TOKEN=$HF_TOKEN" >> $GITHUB_ENV
- name: Install the onnxruntime-genai Python wheel and run python test
run: |
echo "Installing the onnxruntime-genai Python wheel and running the Python tests"
docker run \
--gpus all \
--rm \
--volume $GITHUB_WORKSPACE:/ort_genai_src \
-e HF_TOKEN=$HF_TOKEN \
-w /ort_genai_src onnxruntimecudabuildx64 bash -c " \
${{ env.PYTHON_EXECUTABLE }} -m pip install /ort_genai_src/build/cuda/wheel/onnxruntime_genai*manylinux*.whl --user && \
${{ env.PYTHON_EXECUTABLE }} -m pip install -r test/python/requirements-cuda.txt --user && \
${{ env.PYTHON_EXECUTABLE }} test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/test_models"
- name: Docker -- Run unit tests
run: |
echo "Running docker image onnxruntimecudabuildx64"
docker run \
--gpus all \
--rm \
--volume $GITHUB_WORKSPACE:/ort_genai_src \
-w /ort_genai_src onnxruntimecudabuildx64 bash -c "/ort_genai_src/build/cuda/test/unit_tests"