saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-06-07 00:13:17 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Update dockerfile readme (#9241)

Browse source 
* Update dockerfiles page

* Delete Dockerfile.server

* Delete Dockerfile.training
This commit is contained in:
Faith Xu 2021-10-01 17:28:26 -07:00 committed by GitHub
parent 11a391a88f
commit 9fe09cb72a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 17 additions and 317 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

38
dockerfiles/Dockerfile.server
View file

@ -1,38 +0,0 @@
#-------------------------------------------------------------------------
# Copyright (c) Microsoft Corporation. All rights reserved.
# Licensed under the MIT License.
#--------------------------------------------------------------------------
# Official docker container for ONNX Runtime Server
# Ubuntu 16.04, CPU version, Python 3.
#--------------------------------------------------------------------------
FROM ubuntu:16.04 AS minimal
MAINTAINER Harry Yang "huayang@microsoft.com"
FROM ubuntu:16.04 AS build
ARG PYTHON_VERSION=3.5
ARG ONNXRUNTIME_REPO=https://github.com/Microsoft/onnxruntime
ARG ONNXRUNTIME_SERVER_BRANCH=master
ENV DEBIAN_FRONTEND noninteractive
RUN apt-get update && \
apt-get install -y sudo git bash
ENV PATH="/opt/cmake/bin:${PATH}"
RUN git clone --single-branch --branch ${ONNXRUNTIME_SERVER_BRANCH} --recursive ${ONNXRUNTIME_REPO} onnxruntime
RUN /onnxruntime/tools/ci_build/github/linux/docker/scripts/install_ubuntu.sh -p ${PYTHON_VERSION} \
&& /onnxruntime/tools/ci_build/github/linux/docker/scripts/install_server_deps.sh
ENV PATH="/usr/local/go/bin:${PATH}"
WORKDIR /
RUN mkdir -p /onnxruntime/build && cd /onnxruntime/build && cmake -DCMAKE_BUILD_TYPE=Release /onnxruntime/server \
&& make -j$(getconf _NPROCESSORS_ONLN)
FROM minimal AS final
COPY --from=build /onnxruntime/build/onnxruntime_server /onnxruntime/server/
COPY --from=build /usr/lib/libonnxruntime.so.1.2.0 /usr/local/lib/libcares.so.2.3.0 /usr/lib/
RUN ln -s /usr/lib/libonnxruntime.so.1.2.0 /usr/lib/libonnxruntime.so && ln -s /usr/local/lib/libcares.so.2.3.0 /usr/local/lib/libcares.so.2 && ln -s /usr/local/lib/libcares.so.2 /usr/local/lib/libcares.so && ldconfig /usr/local/lib && apt-get update \
&& apt-get install -y libgomp1 libre2-1v5 libssl1.0.0
WORKDIR /onnxruntime/server/
ENTRYPOINT ["/onnxruntime/server/onnxruntime_server"]

224
dockerfiles/Dockerfile.training
View file

@ -1,224 +0,0 @@
# Copyright (c) Microsoft Corporation. All rights reserved.
# Licensed under the MIT License.
# multi-stage arguments (repeat ARG NAME below)
ARG UCX_VERSION=1.8.0
ARG OPENMPI_VERSION=4.0.4
ARG CONDA_VERSION=4.7.10
ARG NUMPY_VERSION=1.18.5
ARG ONNX_VERSION=1.7.0
ARG PYTORCH_VERSION=1.6.0
ARG BUILD_CONFIG=Release
ARG OPENMPI_PATH=/opt/openmpi-${OPENMPI_VERSION}
ARG COMMIT=master
# cuda development image for building sources
FROM nvcr.io/nvidia/cuda:11.1.1-cudnn8-devel-ubuntu18.04 as builder
# set location for builds
WORKDIR /stage
# install curl, git, ssh (required by MPI when running ORT tests)
RUN apt-get -y update &&\
apt-get -y --no-install-recommends install \
curl \
git \
language-pack-en \
openssh-client \
unattended-upgrades
# update existing packages to minimize security vulnerabilities
RUN unattended-upgrade
RUN locale-gen en_US.UTF-8 && \
update-locale LANG=en_US.UTF-8 && \
curl -O -L https://github.com/Kitware/CMake/releases/download/v3.21.0/cmake-3.21.0-linux-x86_64.tar.gz && \
tar -zxf cmake-3.21.0-linux-x86_64.tar.gz --strip=1 -C /usr && \
rm -rf cmake-3.21.0-linux-x86_64.tar.gz
# install miniconda (comes with python 3.7 default)
ARG CONDA_VERSION
ARG CONDA_URL=https://repo.anaconda.com/miniconda/Miniconda3-${CONDA_VERSION}-Linux-x86_64.sh
RUN cd /stage && curl -fSsL --insecure ${CONDA_URL} -o install-conda.sh &&\
/bin/bash ./install-conda.sh -b -p /opt/conda &&\
/opt/conda/bin/conda clean -ya
ENV PATH=/opt/conda/bin:${PATH}
# install setuptools, numpy, and onnx
ARG NUMPY_VERSION
ARG ONNX_VERSION
RUN conda install -y \
setuptools \
numpy=${NUMPY_VERSION} &&\
pip install \
onnx=="${ONNX_VERSION}"
# install cerberus for the new pytorch front-end
RUN pip install cerberus
# build ucx suite
# note: openmpi will not select ucx without multithreading enabled
ARG UCX_VERSION
ARG UCX_TARNAME=ucx-$UCX_VERSION
ARG UCX_URL=https://github.com/openucx/ucx/releases/download/v${UCX_VERSION}/${UCX_TARNAME}.tar.gz
RUN apt-get -y update && apt-get -y --no-install-recommends install \
libibverbs-dev \
libnuma-dev &&\
cd /stage && curl -fSsL ${UCX_URL} | tar xzf - &&\
cd ${UCX_TARNAME} &&\
./configure \
--prefix=/opt/ucx \
--with-cuda=/usr/local/cuda \
--with-verbs=/usr/lib/x86_64-linux-gnu \
--enable-mt &&\
make -j"$(nproc)" &&\
make install
# build openmpi (use --prefix /opt/openmpi-xxx to move to runtime image)
# note: require --enable-orterun-prefix-by-default for Azure machine learning compute
# note: disable verbs as we use ucx middleware and don't want btl openib warnings
ARG OPENMPI_VERSION
ARG OPENMPI_PATH
ARG OPENMPI_TARNAME=openmpi-${OPENMPI_VERSION}
ARG OPENMPI_URL=https://download.open-mpi.org/release/open-mpi/v%OMPI_BASE%/${OPENMPI_TARNAME}.tar.gz
RUN export OMPI_BASE=${OPENMPI_VERSION%.*} &&\
cd /stage && curl -fSsL `echo ${OPENMPI_URL} | sed s/%OMPI_BASE%/$OMPI_BASE/` | tar xzf - &&\
cd ${OPENMPI_TARNAME} &&\
./configure \
--prefix=${OPENMPI_PATH} \
--with-ucx=/opt/ucx \
--without-verbs \
--with-cuda=/usr/local/cuda \
--enable-mpirun-prefix-by-default \
--enable-orterun-prefix-by-default \
--enable-mca-no-build=btl-uct &&\
make -j"$(nproc)" install &&\
ldconfig
ENV PATH=${OPENMPI_PATH}/bin:$PATH
ENV LD_LIBRARY_PATH=${OPENMPI_PATH}/lib:$LD_LIBRARY_PATH
# install mpi4py (be sure to link existing /opt/openmpi-xxx)
RUN CC=mpicc MPICC=mpicc pip install mpi4py --no-binary mpi4py
# install pytorch
ARG PYTORCH_VERSION
RUN pip install torch==${PYTORCH_VERSION}
# in case you need to build pytorch:
# note: if you want specific branch or to link system cuda libraries or MPI
# note: recommend using many high-frequency cores (e.g. 32+ skylake cores)
# ENV CUDA_HOME="/usr/local/cuda" \
# CUDNN_LIBRARY="/usr/lib/x86_64-linux-gnu" \
# NCCL_INCLUDE_DIR="/usr/include" \
# NCCL_LIB_DIR="/usr/lib/x86_64-linux-gnu" \
# USE_SYSTEM_NCCL=1
# RUN conda install -y \
# mkl \
# mkl-include \
# ninja \
# pyyaml \
# cffi &&\
# cd /stage && git clone https://github.com/pytorch/pytorch.git &&\
# cd pytorch &&\
# git checkout v1.6.0 &&\
# git submodule update --init --recursive &&\
# python setup.py bdist_wheel -d build/wheel &&\
# pip install build/wheel/*.whl
# build onnxruntime wheel with cuda and mpi support
ARG BUILD_CONFIG
ARG COMMIT
RUN cd /stage && git clone https://github.com/microsoft/onnxruntime.git &&\
cd onnxruntime &&\
git checkout ${COMMIT} &&\
cp ThirdPartyNotices.txt /stage/ThirdPartyNotices.txt &&\
cp dockerfiles/LICENSE-IMAGE.txt /stage/LICENSE-IMAGE.txt &&\
python tools/ci_build/build.py \
--cmake_extra_defines \
ONNXRUNTIME_VERSION=`cat ./VERSION_NUMBER` \
--config ${BUILD_CONFIG} \
--enable_training \
--mpi_home ${OPENMPI_PATH} \
--use_cuda \
--cuda_home /usr/local/cuda \
--cudnn_home /usr/lib/x86_64-linux-gnu/ \
--nccl_home /usr/lib/x86_64-linux-gnu/ \
--update \
--parallel \
--build_dir build \
--build \
--build_wheel \
--skip_tests --cmake_extra_defines 'CMAKE_CUDA_ARCHITECTURES=35;37;50;52;60;61;70;75;80;86' &&\
pip install build/${BUILD_CONFIG}/dist/*.whl
# Install AzureML support and commonly used packages.
RUN pip install azureml-defaults sentencepiece==0.1.92 transformers==2.11.0 msgpack==1.0.0 tensorboardX==1.8 tensorboard==2.3.0
# switch to cuda runtime environment
# note: launch with --gpus all or nvidia-docker
FROM nvcr.io/nvidia/cuda:11.1.1-cudnn8-runtime-ubuntu18.04
WORKDIR /stage
# install ucx
# note: launch with --cap-add=sys_nice to avoid 'mbind' warnings
COPY --from=builder /opt/ucx /opt/ucx
ENV PATH=/opt/ucx/bin:$PATH
ENV LD_LIBRARY_PATH=/opt/ucx/lib:$LD_LIBRARY_PATH
# install openmpi
# note: permit mpirun as root for Azure cluster submissions
# note: enforce openmpi select ucx or fail
ARG OPENMPI_VERSION
ARG OPENMPI_PATH
COPY --from=builder ${OPENMPI_PATH} ${OPENMPI_PATH}
ENV PATH=${OPENMPI_PATH}/bin:$PATH
ENV LD_LIBRARY_PATH=${OPENMPI_PATH}/lib:$LD_LIBRARY_PATH
ENV OMPI_ALLOW_RUN_AS_ROOT=1
ENV OMPI_ALLOW_RUN_AS_ROOT_CONFIRM=1
ENV OMPI_MCA_pml=ucx
RUN apt-get -y update && apt-get -y --no-install-recommends install \
openssh-server \
openssh-client \
libibverbs-dev \
libnuma-dev &&\
ldconfig
# copy conda environment (includes numpy, mpi4py, pytorch, onnxruntime)
COPY --from=builder /opt/conda /opt/conda
ENV PATH=/opt/conda/bin:${PATH}
# make ssh/sshd less strict for wiring containers on Azure VM scale set
# note: use 'service ssh start' to launch sshd (will fail if 22 in use)
# note: can also set port != 22 and set port=X in MPI hosts file
# note: need to setup password free ssh login between MPI hosts
RUN sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/g' \
/etc/ssh/sshd_config &&\
sed -i 's/#StrictModes yes/StrictModes no/g' \
/etc/ssh/sshd_config &&\
sed -i 's/# StrictHostKeyChecking ask/ StrictHostKeyChecking no/g' \
/etc/ssh/ssh_config &&\
mkdir /run/sshd &&\
chmod u=rwx,go=rx /run/sshd
# export versions
ARG UCX_VERSION
ARG OPENMPI_VERSION
ARG CONDA_VERSION
ARG NUMPY_VERSION
ARG ONNX_VERSION
ARG PYTORCH_VERSION
LABEL UCX_VERSION=${UCX_VERSION}
LABEL OPENMPI_VERSION=${OPENMPI_VERSION}
LABEL CONDA_VERSION=${CONDA_VERSION}
LABEL NUMPY_VERSION=${NUMPY_VERSION}
LABEL ONNX_VERSION=${ONNX_VERSION}
LABEL PYTORCH_VERSION=${PYTORCH_VERSION}
# clean\finalize environment
# note: adds onnxruntime license and third party notices
RUN apt-get purge -y build-essential &&\
apt-get autoremove -y &&\
rm -fr /stage
WORKDIR /workspace
COPY --from=builder /stage/*.txt /workspace/

72
dockerfiles/README.md
View file

@ -1,41 +1,25 @@
# Docker Containers for ONNX Runtime
**Dockerfiles**
- CPU [Dockerfile](Dockerfile.source), [Instructions](#cpu)
- CUDA + CUDNN: [Dockerfile](Dockerfile.cuda), [Instructions](#cuda)
- TensorRT: [Dockerfile](Dockerfile.tensorrt), [Instructions](#tensorrt)
- OpenVINO: [Dockerfile](Dockerfile.openvino), [Instructions](#openvino)
- Nuphar: [Dockerfile](Dockerfile.nuphar), [Instructions](#nuphar)
- ARM 32v7: [Dockerfile](Dockerfile.arm32v7), [Instructions](#arm-32v7)
- NVIDIA Jetson TX1/TX2/Nano/Xavier: [Dockerfile](Dockerfile.jetson), [Instructions](#nvidia-jetson-tx1tx2nanoxavier)
- ONNX-Ecosystem (CPU + Converters): [Dockerfile](https://github.com/onnx/onnx-docker/blob/master/onnx-ecosystem/Dockerfile), [Instructions](https://github.com/onnx/onnx-docker/tree/master/onnx-ecosystem)
- ONNX Runtime Server: [Dockerfile](Dockerfile.server), [Instructions](#onnx-runtime-server)
# Dockerfiles
**Execution Providers**
- CPU: [Dockerfile](Dockerfile.source), [Instructions](#cpu)
- CUDA/cuDNN: [Dockerfile](Dockerfile.cuda), [Instructions](#cuda)
- MIGraphX: [Dockerfile](Dockerfile.migraphx), [Instructions](#migraphx)
- NUPHAR: [Dockerfile](Dockerfile.nuphar), [Instructions](#nuphar)
- OpenVINO: [Dockerfile](Dockerfile.openvino), [Instructions](#openvino)
- TensorRT: [Dockerfile](Dockerfile.tensorrt), [Instructions](#tensorrt)
- VitisAI: [Dockerfile](Dockerfile.vitisai)
**Published Microsoft Container Registry (MCR) Images**
**Platforms**
- ARM 32v7: [Dockerfile](Dockerfile.arm32v7), [Instructions](#arm-3264)
- ARM 64: [Dockerfile](Dockerfile.arm64), [Instructions](#arm-3264)
- NVIDIA Jetson TX1/TX2/Nano/Xavier: [Dockerfile](Dockerfile.jetson), [Instructions](#nvidia-jetson-tx1tx2nanoxavier)
Use `docker pull` with any of the images and tags below to pull an image and try for yourself. Note that the CPU and CUDA images include additional dependencies like miniconda for compatibility with AzureML image deployment.
**Other**
- ORT Training (torch-ort): [Dockerfiles](https://github.com/pytorch/ort/tree/main/docker)
- ONNX-Ecosystem (CPU + Converters): [Dockerfile](https://github.com/onnx/onnx-docker/blob/master/onnx-ecosystem/Dockerfile), [Instructions](https://github.com/onnx/onnx-docker/tree/master/onnx-ecosystem)
**Example**: Run `docker pull mcr.microsoft.com/azureml/onnxruntime:latest-cuda` to pull the latest released docker image with ONNX Runtime GPU, CUDA, and CUDNN support.
| Build Flavor | Base Image | ONNX Runtime Docker Image tags | Latest |
|-------------------|---------------------------------------|---------------------------------------|-------------------------|
| Source (CPU) | mcr.microsoft.com/azureml/onnxruntime | :v0.4.0, :v0.5.0, v0.5.1, :v1.0.0, :v1.2.0, :v1.3.0, :v1.4.0, :v1.5.2 | :latest |
| CUDA (GPU) | mcr.microsoft.com/azureml/onnxruntime | :v0.4.0-cuda10.0-cudnn7, :v0.5.0-cuda10.1-cudnn7, :v0.5.1-cuda10.1-cudnn7, :v1.0.0-cuda10.1-cudnn7, :v1.2.0-cuda10.1-cudnn7, :v1.3.0-cuda10.1-cudnn7, :v1.4.0-cuda10.1-cudnn7, :v1.5.2-cuda10.2-cudnn8 | :latest-cuda |
| OpenVino | hub.docker.com/repository/docker/openvino/onnxruntime_ep_ubuntu18 | :2021.3, :2021.4 | :latest |
| OpenVino (VAD-M) | mcr.microsoft.com/azureml/onnxruntime | :v0.5.0-openvino-r1.1-vadm, :v1.0.0-openvino-r1.1-vadm, :v1.4.0-openvino-2020.3.194-vadm, :v1.5.2-openvino-2020.4.287-vadm | :latest-openvino-vadm |
| OpenVino (MYRIAD) | mcr.microsoft.com/azureml/onnxruntime | :v0.5.0-openvino-r1.1-myriad, :v1.0.0-openvino-r1.1-myriad, :v1.3.0-openvino-2020.2.120-myriad, :v1.4.0-openvino-2020.3.194-myriad, :v1.5.2-openvino-2020.4.287-myriad | :latest-openvino-myriad |
| OpenVino (CPU) | mcr.microsoft.com/azureml/onnxruntime | :v1.0.0-openvino-r1.1-cpu, :v1.3.0-openvino-2020.2.120-cpu, :v1.4.0-openvino-2020.3.194-cpu, :v1.5.2-openvino-2020.4.287-cpu | :latest-openvino-cpu |
| OpenVINO (GPU) | mcr.microsoft.com/azureml/onnxruntime | :v1.3.0-openvino-2020.2.120-gpu, :v1.4.0-openvino-2020.3.194-gpu, :v1.5.2-openvino-2020.4.287-gpu | :latest-openvino-gpu|
| Nuphar | mcr.microsoft.com/azureml/onnxruntime | | :latest-nuphar |
| Server | mcr.microsoft.com/onnxruntime/server | :v0.4.0, :v0.5.0, :v0.5.1, :v1.0.0 | :latest |
| MIGraphX (GPU) | mcr.microsoft.com/azureml/onnxruntime | :v0.6 | :latest |
| Training ([usage](https://github.com/microsoft/onnxruntime-training-examples))| mcr.microsoft.com/azureml/onnxruntime-training | :0.1-rc1-openmpi4.0-cuda10.1-cudnn7.6-nccl2.4.8, :0.1-rc2-openmpi4.0-cuda10.2-cudnn7.6-nccl2.7.6, :0.1-rc3.1-openmpi4.0-cuda10.2-cudnn8.0-nccl2.7 | :latest |
---
# Building and using Docker images
# Instructions
## CPU
**Ubuntu 18.04, CPU, Python Bindings**
@ -282,7 +266,7 @@ Nothing else from ONNX Runtime source tree will be copied/installed to the image
Note: When running the container you built in Docker, please either use 'nvidia-docker' command instead of 'docker', or use Docker command-line options to make sure NVIDIA runtime will be used and appropiate files mounted from host. Otherwise, CUDA libraries won't be found. You can also [set NVIDIA runtime as default in Docker](https://github.com/dusty-nv/jetson-containers#docker-default-runtime).
## Nuphar
## NUPHAR
*Public Preview*
**Ubuntu 16.04, Python Bindings**
@ -316,25 +300,3 @@ git submodule update --init
```
docker run -it --device=/dev/kfd --device=/dev/dri --group-add video onnxruntime-migraphx
```
## ONNX Runtime Server
*Public Preview*
**Ubuntu 16.04**
1. Build the docker image from the Dockerfile in this repository
```
docker build -t {docker_image_name} -f Dockerfile.server .
```
2. Run the ONNXRuntime server with the image created in step 1
```
docker run -v {localModelAbsoluteFolder}:{dockerModelAbsoluteFolder} -p {your_local_port}:8001 {imageName} --model_path {dockerModelAbsolutePath}
```
3. Send HTTP requests to the container running ONNX Runtime Server
Send HTTP requests to the docker container through the binding local port. Here is the full [usage document](../docs/ONNX_Runtime_Server_Usage.md).
```
curl -X POST -d "@request.json" -H "Content-Type: application/json" http://0.0.0.0:{your_local_port}/v1/models/mymodel/versions/3:predict
```