mirror of
https://github.com/saymrwulf/onnxruntime.git
synced 2026-07-28 20:11:22 +00:00
7e211f0e03
Some CI jobs may interrupted unexpectedly and didn't execute umount data step. The data left in host device will cause `device or resource busy` and make subsequent CI jobs fail. Move the mount data step into docker container, the host machine will not be occupied when CI jobs exit incorrectly.
61 lines
1.8 KiB
Docker
61 lines
1.8 KiB
Docker
FROM rocm/pytorch:rocm5.5_ubuntu20.04_py3.8_pytorch_1.13.1
|
|
|
|
ARG BUILD_UID=1001
|
|
ARG BUILD_USER=onnxruntimedev
|
|
RUN adduser --uid $BUILD_UID $BUILD_USER
|
|
RUN echo "$BUILD_USER ALL=(ALL) NOPASSWD:ALL" >> /etc/sudoers.d/$BUILD_USER
|
|
|
|
WORKDIR /home/$BUILD_USER
|
|
|
|
# from rocm/pytorch's image, work around ucx's dlopen replacement conflicting with shared provider
|
|
RUN cd /opt/mpi_install/ucx/build &&\
|
|
make clean &&\
|
|
../contrib/configure-release --prefix=/opt/ucx --without-rocm &&\
|
|
make -j $(nproc) &&\
|
|
make install
|
|
|
|
# CMake
|
|
ENV CMAKE_VERSION=3.26.3
|
|
RUN cd /usr/local && \
|
|
wget -q -O - https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-Linux-x86_64.tar.gz | tar zxf -
|
|
ENV PATH=/usr/local/cmake-${CMAKE_VERSION}-linux-x86_64/bin:${PATH}
|
|
|
|
# ccache
|
|
RUN mkdir -p /tmp/ccache && \
|
|
cd /tmp/ccache && \
|
|
wget -q -O - https://github.com/ccache/ccache/releases/download/v4.7.4/ccache-4.7.4-linux-x86_64.tar.xz | tar --strip 1 -J -xf - && \
|
|
cp /tmp/ccache/ccache /usr/bin && \
|
|
rm -rf /tmp/ccache
|
|
|
|
RUN apt-get update && apt-get install -y cifs-utils
|
|
|
|
USER $BUILD_USER
|
|
|
|
# rocm-ci branch contains instrumentation needed for loss curves and perf
|
|
RUN git clone https://github.com/microsoft/huggingface-transformers.git &&\
|
|
cd huggingface-transformers &&\
|
|
git checkout rocm-ci &&\
|
|
pip install -e .
|
|
|
|
RUN pip install \
|
|
numpy==1.24.1 \
|
|
onnx \
|
|
cerberus \
|
|
sympy \
|
|
h5py \
|
|
datasets==1.9.0 \
|
|
requests \
|
|
sacrebleu==1.5.1 \
|
|
sacremoses \
|
|
scipy==1.10.0 \
|
|
scikit-learn \
|
|
tokenizers \
|
|
sentencepiece \
|
|
dill==0.3.4 \
|
|
wget \
|
|
pytorch_lightning==1.6.0 \
|
|
pytest-xdist \
|
|
pytest-rerunfailures
|
|
|
|
RUN pip install torch-ort --no-dependencies
|
|
ENV ORTMODULE_ONNX_OPSET_VERSION=15
|