Skip to content

Commit

Permalink
Remove onnxruntime build from build.py, now in onnxruntime_backend re…
Browse files Browse the repository at this point in the history 
…po (#2481)

* Remove onnxruntime build from build.py, now in onnxruntime_backend repo

* Review edits
  • Loading branch information
deadeyegoodwin committed Feb 4, 2021
1 parent 3ae24b5 commit 470bb89
Showing 1 changed file with 21 additions and 254 deletions.
275 changes: 21 additions & 254 deletions build.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -327,12 +327,15 @@ def pytorch_cmake_args(images):
def onnxruntime_cmake_args():
cargs = [
'-DTRITON_ENABLE_ONNXRUNTIME_TENSORRT=ON',
'-DTRITON_ONNXRUNTIME_INCLUDE_PATHS=/opt/tritonserver/include/onnxruntime',
'-DTRITON_ONNXRUNTIME_LIB_PATHS=/opt/tritonserver/backends/onnxruntime'
'-DTRITON_BUILD_ONNXRUNTIME_VERSION={}'.format(TRITON_VERSION_MAP[FLAGS.version][2]),
'-DTRITON_BUILD_CONTAINER_VERSION={}'.format(TRITON_VERSION_MAP[FLAGS.version][1])
]

if TRITON_VERSION_MAP[FLAGS.version][3] is not None:
cargs.append('-DTRITON_ENABLE_ONNXRUNTIME_OPENVINO=ON')
cargs.append('-DTRITON_BUILD_ONNXRUNTIME_OPENVINO_VERSION={}'.format(
TRITON_VERSION_MAP[FLAGS.version][3]))

return cargs


Expand Down Expand Up @@ -372,135 +375,13 @@ def dali_cmake_args():

def create_dockerfile_buildbase(ddir, dockerfile_name, argmap, backends):
df = '''
#
# Multistage build.
#
ARG TRITON_VERSION={}
ARG TRITON_CONTAINER_VERSION={}
ARG BASE_IMAGE={}
'''.format(argmap['TRITON_VERSION'], argmap['TRITON_CONTAINER_VERSION'],
argmap['BASE_IMAGE'])
if 'onnxruntime' in backends:
df += '''
ARG ONNX_RUNTIME_VERSION={}
'''.format(argmap['ONNX_RUNTIME_VERSION'])
if 'ONNX_RUNTIME_OPENVINO_VERSION' in argmap:
df += '''
ARG ONNX_RUNTIME_OPENVINO_VERSION={}
'''.format(argmap['ONNX_RUNTIME_OPENVINO_VERSION'])

if 'onnxruntime' in backends:
df += '''
############################################################################
## Onnx Runtime stage: Build Onnx Runtime with up-to-date CUDA, CUDNN
############################################################################
FROM ${BASE_IMAGE} AS tritonserver_onnx
# Onnx Runtime release version from top of file
ARG ONNX_RUNTIME_VERSION
ARG ONNXRUNTIME_REPO=https://github.com/microsoft/onnxruntime
# Ensure apt-get won't prompt for selecting options
ENV DEBIAN_FRONTEND=noninteractive
WORKDIR /workspace
# The Onnx Runtime dockerfile is the collection of steps in
# https://github.com/microsoft/onnxruntime/tree/master/dockerfiles
# Install dependencies from
# onnxruntime/dockerfiles/scripts/install_common_deps.sh. We don't run
# that script directly because we don't want cmake installed from that
# file.
RUN apt-get update && apt-get install -y --no-install-recommends \
wget \
zip \
ca-certificates \
build-essential \
cmake \
curl \
libcurl4-openssl-dev \
libssl-dev \
python3-dev \
python3-pip
RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-4.5.11-Linux-x86_64.sh -O ~/miniconda.sh \
--no-check-certificate && \
/bin/bash ~/miniconda.sh -b -p /opt/miniconda && \
rm ~/miniconda.sh && \
/opt/miniconda/bin/conda clean -ya
# Allow configure to pick up GDK and CuDNN where it expects it.
# (Note: $CUDNN_VERSION is defined by NVidia's base image)
RUN _CUDNN_VERSION=$(echo $CUDNN_VERSION | cut -d. -f1-2) && \
mkdir -p /usr/local/cudnn-$_CUDNN_VERSION/cuda/include && \
ln -s /usr/include/cudnn.h /usr/local/cudnn-$_CUDNN_VERSION/cuda/include/cudnn.h && \
mkdir -p /usr/local/cudnn-$_CUDNN_VERSION/cuda/lib64 && \
ln -s /etc/alternatives/libcudnn_so /usr/local/cudnn-$_CUDNN_VERSION/cuda/lib64/libcudnn.so
'''
if 'ONNX_RUNTIME_OPENVINO_VERSION' in argmap:
df += '''
# Install OpenVINO
ARG ONNX_RUNTIME_OPENVINO_VERSION
ENV INTEL_OPENVINO_DIR /opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110
ENV LD_LIBRARY_PATH $INTEL_OPENVINO_DIR/deployment_tools/inference_engine/lib/intel64:$INTEL_OPENVINO_DIR/deployment_tools/ngraph/lib:$INTEL_OPENVINO_DIR/deployment_tools/inference_engine/external/tbb/lib:/usr/local/openblas/lib:$LD_LIBRARY_PATH
ENV PYTHONPATH $INTEL_OPENVINO_DIR/tools:$PYTHONPATH
ENV IE_PLUGINS_PATH $INTEL_OPENVINO_DIR/deployment_tools/inference_engine/lib/intel64
RUN wget https://apt.repos.intel.com/openvino/2021/GPG-PUB-KEY-INTEL-OPENVINO-2021 && \
apt-key add GPG-PUB-KEY-INTEL-OPENVINO-2021 && rm GPG-PUB-KEY-INTEL-OPENVINO-2021 && \
cd /etc/apt/sources.list.d && \
echo "deb https://apt.repos.intel.com/openvino/2021 all main">intel-openvino-2021.list && \
apt update && \
apt install -y intel-openvino-dev-ubuntu20-${ONNX_RUNTIME_OPENVINO_VERSION}.110 && \
cd ${INTEL_OPENVINO_DIR}/install_dependencies && ./install_openvino_dependencies.sh
ARG INTEL_COMPUTE_RUNTIME_URL=https://github.com/intel/compute-runtime/releases/download/19.41.14441
RUN wget ${INTEL_COMPUTE_RUNTIME_URL}/intel-gmmlib_19.3.2_amd64.deb && \
wget ${INTEL_COMPUTE_RUNTIME_URL}/intel-igc-core_1.0.2597_amd64.deb && \
wget ${INTEL_COMPUTE_RUNTIME_URL}/intel-igc-opencl_1.0.2597_amd64.deb && \
wget ${INTEL_COMPUTE_RUNTIME_URL}/intel-opencl_19.41.14441_amd64.deb && \
wget ${INTEL_COMPUTE_RUNTIME_URL}/intel-ocloc_19.41.14441_amd64.deb && \
dpkg -i *.deb && rm -rf *.deb
'''
df += '''
# ONNX Runtime
RUN git clone -b rel-${ONNX_RUNTIME_VERSION} --recursive ${ONNXRUNTIME_REPO} onnxruntime && \
(cd onnxruntime && git submodule update --init --recursive)
# Need to patch until https://github.com/onnx/onnx-tensorrt/pull/568
# is merged and used in ORT
COPY build/onnxruntime/onnx_tensorrt.patch /tmp/onnx_tensorrt.patch
RUN cd /workspace/onnxruntime/cmake/external/onnx-tensorrt && \
patch -i /tmp/onnx_tensorrt.patch builtin_op_importers.cpp
ARG COMMON_BUILD_ARGS="--skip_submodule_sync --parallel --build_shared_lib --use_openmp"
RUN mkdir -p /workspace/build
'''
extra_build_args = ''
if 'ONNX_RUNTIME_OPENVINO_VERSION' in argmap:
extra_build_args += '--use_openvino CPU_FP32'
df += '''
RUN python3 /workspace/onnxruntime/tools/ci_build/build.py --build_dir /workspace/build \
--config Release $COMMON_BUILD_ARGS \
--cmake_extra_defines ONNXRUNTIME_VERSION=$(cat /workspace/onnxruntime/VERSION_NUMBER) \
--cuda_home /usr/local/cuda \
--cudnn_home /usr/local/cudnn-$(echo $CUDNN_VERSION | cut -d. -f1-2)/cuda \
--tensorrt_home /usr/src/tensorrt \
--use_cuda \
--use_tensorrt \
--update \
--build {}
'''.format(extra_build_args)
df += '''
# Record version of ONNX used for ORT compilation,
RUN cat /workspace/onnxruntime/cmake/external/onnx/VERSION_NUMBER > /workspace/ort_onnx_version.txt
'''

df += '''
############################################################################
## Final stage: Install and arrange all dependencies needed for build
############################################################################
FROM ${BASE_IMAGE}
ARG TRITON_VERSION
Expand Down Expand Up @@ -560,89 +441,6 @@ def create_dockerfile_buildbase(ddir, dockerfile_name, argmap, backends):
apt-get install -y --no-install-recommends \
cmake-data=3.18.4-0kitware1ubuntu20.04.1 cmake=3.18.4-0kitware1ubuntu20.04.1
'''
if 'onnxruntime' in backends:
df += '''
# Onnx Runtime headers and library
# Put include files to same directory as ONNX Runtime changed the include path
# https://github.com/microsoft/onnxruntime/pull/1461
ARG ONNX_RUNTIME_VERSION
COPY --from=tritonserver_onnx /workspace/onnxruntime/include/onnxruntime/core/session/onnxruntime_c_api.h \
/opt/tritonserver/include/onnxruntime/
COPY --from=tritonserver_onnx /workspace/onnxruntime/include/onnxruntime/core/session/onnxruntime_session_options_config_keys.h \
/opt/tritonserver/include/onnxruntime/
COPY --from=tritonserver_onnx /workspace/onnxruntime/include/onnxruntime/core/providers/cpu/cpu_provider_factory.h \
/opt/tritonserver/include/onnxruntime/
COPY --from=tritonserver_onnx /workspace/onnxruntime/include/onnxruntime/core/providers/cuda/cuda_provider_factory.h \
/opt/tritonserver/include/onnxruntime/
COPY --from=tritonserver_onnx /workspace/onnxruntime/include/onnxruntime/core/providers/tensorrt/tensorrt_provider_factory.h \
/opt/tritonserver/include/onnxruntime/
'''
if 'ONNX_RUNTIME_OPENVINO_VERSION' in argmap:
df += '''
ARG ONNX_RUNTIME_OPENVINO_VERSION
COPY --from=tritonserver_onnx /workspace/onnxruntime/include/onnxruntime/core/providers/openvino/openvino_provider_factory.h \
/opt/tritonserver/include/onnxruntime/
COPY --from=tritonserver_onnx /workspace/build/Release/libonnxruntime_providers_openvino.so \
/opt/tritonserver/backends/onnxruntime/
'''
df += '''
COPY --from=tritonserver_onnx /workspace/build/Release/libonnxruntime.so.${ONNX_RUNTIME_VERSION} \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx /workspace/build/Release/libonnxruntime_providers_shared.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx /workspace/build/Release/libonnxruntime_providers_tensorrt.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx /workspace/onnxruntime/LICENSE \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx /workspace/ort_onnx_version.txt \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx /workspace/build/Release/onnxruntime_perf_test \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx /workspace/build/Release/onnx_test_runner \
/opt/tritonserver/backends/onnxruntime/
RUN cd /opt/tritonserver/backends/onnxruntime && \
ln -sf libonnxruntime.so.${ONNX_RUNTIME_VERSION} libonnxruntime.so
'''
if 'ONNX_RUNTIME_OPENVINO_VERSION' in argmap:
df += '''
# Minimum OpenVINO libraries required by ONNX Runtime to link and to run
# with OpenVINO Execution Provider
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/inference_engine/lib/intel64/libinference_engine.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/inference_engine/lib/intel64/libinference_engine_legacy.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/inference_engine/lib/intel64/libinference_engine_transformations.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/ngraph/lib/libngraph.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/ngraph/lib/libonnx_importer.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/inference_engine/external/tbb/lib/libtbb.so.2 \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/inference_engine/lib/intel64/plugins.xml \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/inference_engine/lib/intel64/libMKLDNNPlugin.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/deployment_tools/inference_engine/lib/intel64/libinference_engine_lp_transformations.so \
/opt/tritonserver/backends/onnxruntime/
COPY --from=tritonserver_onnx \
/opt/intel/openvino_${ONNX_RUNTIME_OPENVINO_VERSION}.110/licensing \
/opt/tritonserver/backends/onnxruntime/LICENSE.openvino
RUN cd /opt/tritonserver/backends/onnxruntime && \
ln -sf libtbb.so.2 libtbb.so && \
for i in `find . -mindepth 1 -maxdepth 1 -type f -name '*\.so*'`; do \
patchelf --set-rpath '$ORIGIN' $i; \
done
'''

# Copy in the triton source. We remove existing contents first in
# case the FROM container has something there already. On windows
Expand All @@ -668,14 +466,6 @@ def create_dockerfile_buildbase(ddir, dockerfile_name, argmap, backends):
ENTRYPOINT []
'''

if 'onnxruntime' in backends:
df += '''
# Copy ONNX custom op library and model (Needed for testing)
COPY --from=tritonserver_onnx /workspace/build/Release/libcustom_op_library.so \
/workspace/qa/L0_custom_ops/
COPY --from=tritonserver_onnx /workspace/build/Release/testdata/custom_op_library/custom_op_test.onnx \
/workspace/qa/L0_custom_ops/
'''
df += '''
ENV TRITON_SERVER_VERSION ${TRITON_VERSION}
ENV NVIDIA_TRITON_SERVER_VERSION ${TRITON_CONTAINER_VERSION}
Expand All @@ -685,12 +475,22 @@ def create_dockerfile_buildbase(ddir, dockerfile_name, argmap, backends):
dfile.write(df)


def create_dockerfile_build(ddir, dockerfile_name, argmap):
def create_dockerfile_build(ddir, dockerfile_name, argmap, backends):
df = '''
FROM tritonserver_builder_image AS build
FROM tritonserver_buildbase
COPY --from=build /tmp/tritonbuild /tmp/tritonbuild
'''

if 'onnxruntime' in backends:
df += '''
# Copy ONNX custom op library and model (needed for testing)
RUN if [ -d /tmp/tritonbuild/onnxruntime ]; then \
cp /tmp/tritonbuild/onnxruntime/install/test/libcustom_op_library.so /workspace/qa/L0_custom_ops/.; \
cp /tmp/tritonbuild/onnxruntime/install/test/custom_op_test.onnx /workspace/qa/L0_custom_ops/.; \
fi
'''

mkdir(ddir)
with open(os.path.join(ddir, dockerfile_name), "w") as dfile:
dfile.write(df)
Expand Down Expand Up @@ -778,10 +578,6 @@ def create_dockerfile(ddir, dockerfile_name, argmap, backends):
COPY --chown=1000:1000 TRITON_VERSION .
COPY --chown=1000:1000 --from=tritonserver_build /tmp/tritonbuild/install/bin/tritonserver bin/
COPY --chown=1000:1000 --from=tritonserver_build /tmp/tritonbuild/install/lib/libtritonserver.so lib/
'''
if 'onnxruntime' in backends:
df += '''
COPY --chown=1000:1000 --from=tritonserver_build /opt/tritonserver/backends/onnxruntime backends/onnxruntime
'''

for noncore in NONCORE_BACKENDS:
Expand All @@ -791,11 +587,6 @@ def create_dockerfile(ddir, dockerfile_name, argmap, backends):
'''
break

if 'onnxruntime' in backends:
df += '''
# Get ONNX version supported
RUN export ONNX_VERSION=`cat backends/onnxruntime/ort_onnx_version.txt`
'''
df += '''
# Extra defensive wiring for CUDA Compat lib
RUN ln -sf ${{_CUDA_COMPAT_PATH}}/lib.real ${{_CUDA_COMPAT_PATH}}/lib \
Expand Down Expand Up @@ -848,22 +639,9 @@ def container_build(backends, images):
base_image,
}

# If building the ORT backend then need to include ORT in
# buildbase image.
if 'onnxruntime' in backends:
if FLAGS.version not in TRITON_VERSION_MAP:
fail('ONNX Runtime version not known for {}'.format(FLAGS.version))
dockerfileargmap['ONNX_RUNTIME_VERSION'] = TRITON_VERSION_MAP[
FLAGS.version][2]
if TRITON_VERSION_MAP[FLAGS.version][3] is not None:
dockerfileargmap[
'ONNX_RUNTIME_OPENVINO_VERSION'] = TRITON_VERSION_MAP[
FLAGS.version][3]

cachefrommap = [
'tritonserver_onnx', 'tritonserver_onnx_cache0',
'tritonserver_onnx_cache1', 'tritonserver_buildbase',
'tritonserver_buildbase_cache0', 'tritonserver_buildbase_cache1'
'tritonserver_buildbase', 'tritonserver_buildbase_cache0',
'tritonserver_buildbase_cache1'
]

cachefromargs = ['--cache-from={}'.format(k) for k in cachefrommap]
Expand All @@ -880,19 +658,8 @@ def container_build(backends, images):
create_dockerfile_buildbase(FLAGS.build_dir, 'Dockerfile.buildbase',
dockerfileargmap, backends)
try:
# First build Dockerfile.buildbase. Because of the way Docker
# does caching with multi-stage images, we must build each
# stage separately to make sure it is cached.

# ONNX Runtime
if 'onnxruntime' in backends:
p = subprocess.Popen(commonargs + cachefromargs + [
'-t', 'tritonserver_onnx', '--target', 'tritonserver_onnx', '.'
])
p.wait()
fail_if(p.returncode != 0, 'docker build tritonserver_onnx failed')

# Final buildbase image
# Create buildbase image, this is an image with all
# dependencies needed for the build.
p = subprocess.Popen(commonargs + cachefromargs +
['-t', 'tritonserver_buildbase', '.'])
p.wait()
Expand Down Expand Up @@ -999,7 +766,7 @@ def container_build(backends, images):
container.remove(force=True)

create_dockerfile_build(FLAGS.build_dir, 'Dockerfile.build',
dockerfileargmap)
dockerfileargmap, backends)
p = subprocess.Popen([
'docker', 'build', '-t', 'tritonserver_build', '-f',
os.path.join(FLAGS.build_dir, 'Dockerfile.build'), '.'
Expand Down

0 comments on commit 470bb89

Please sign in to comment.