[Serving] add fastdeployserver dockerfile for cuda11.2 (PaddlePaddle#…

…1169) * add fastdeployserver dockerfile for cuda11.2 * add docks * update --------- Co-authored-by: heliqi <[email protected]>
leyond123 · Jan 30, 2023 · 8c651f9 · 8c651f9
1 parent 84d41dc
commit 8c651f9
Show file tree

Hide file tree

Showing 4 changed files with 164 additions and 0 deletions.
diff --git a/serving/Dockerfile_CUDA_11_2 b/serving/Dockerfile_CUDA_11_2
@@ -0,0 +1,83 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+FROM nvidia/cuda:11.2.2-cudnn8-devel-ubuntu20.04
+
+ARG http_proxy
+ARG https_proxy
+
+#Install the build dependencies
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive TZ=Asia/Shanghai apt-get install -y --no-install-recommends curl wget vim git patchelf python3-dev python3-pip \
+    python3-setuptools build-essential libgl1-mesa-glx libglib2.0-dev ca-certificates libb64-dev datacenter-gpu-manager \
+    libssl-dev zlib1g-dev rapidjson-dev libboost-dev libre2-dev librdmacm-dev libnuma-dev libarchive-dev unzip && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+RUN ln -s /usr/bin/python3 /usr/bin/python;
+RUN pip3 install --upgrade pip
+
+# install cmake
+WORKDIR /home
+RUN wget -q https://github.com/Kitware/CMake/releases/download/v3.18.6/cmake-3.18.6-Linux-x86_64.tar.gz && tar -zxvf cmake-3.18.6-Linux-x86_64.tar.gz
+ENV PATH=/home/cmake-3.18.6-Linux-x86_64/bin:$PATH
+
+
+#install triton
+ENV TAG=r21.10
+RUN git clone https://github.com/triton-inference-server/server.git -b $TAG && \
+   cd server && \
+   mkdir -p build/tritonserver/install && \
+   python3 build.py \
+     --build-dir `pwd`/build \
+     --no-container-build \
+     --backend=ensemble \
+     --enable-gpu \
+     --endpoint=grpc \
+     --endpoint=http \
+     --enable-stats \
+     --enable-tracing \
+     --enable-logging \
+     --enable-stats \
+     --enable-metrics \
+     --enable-gpu-metrics \
+     --cmake-dir `pwd`/build \
+     --repo-tag=common:$TAG \
+     --repo-tag=core:$TAG \
+     --repo-tag=backend:$TAG \
+     --repo-tag=thirdparty:$TAG \
+     --backend=python:$TAG
+
+COPY python/dist/*.whl /opt/fastdeploy/
+RUN python3 -m pip install  /opt/fastdeploy/*.whl \
+    && rm -rf /opt/fastdeploy/*.whl
+
+
+# compile triton-inference-server/server，copy tritonserver and python backend into image
+# triton server
+RUN mkdir -p /opt/tritonserver && cp -r /home/server/build/tritonserver/install/* /opt/tritonserver
+# python backend
+RUN mkdir -p /opt/tritonserver/backends/python && cp -r /home/server/build/python/install/backends/python /opt/tritonserver/backends/
+
+# copy compiled fastdeploy backend into image
+COPY serving/build/libtriton_fastdeploy.so /opt/tritonserver/backends/fastdeploy/
+
+# rename tritonserver to fastdeployserver
+RUN mv /opt/tritonserver/bin/tritonserver /opt/tritonserver/bin/fastdeployserver
+
+# copy compiled fastdeploy_install into image
+COPY build/fastdeploy_install/* /opt/fastdeploy/
+
+# Set environment variable
+ENV LD_LIBRARY_PATH="/opt/fastdeploy/lib:/opt/fastdeploy/third_libs/install/onnxruntime/lib:/opt/fastdeploy/third_libs/install/paddle2onnx/lib:/opt/fastdeploy/third_libs/install/paddle_inference/paddle/lib:/opt/fastdeploy/third_libs/install/openvino/runtime/lib/:/opt/fastdeploy/third_libs/install/tensorrt/lib/:/opt/fastdeploy/third_libs/install/opencv/lib64/:$LD_LIBRARY_PATH"
+ENV PATH="/opt/tritonserver/bin:$PATH"
diff --git a/serving/docs/EN/compile-en.md b/serving/docs/EN/compile-en.md
@@ -18,6 +18,17 @@ cd ../
 docker build -t paddlepaddle/fastdeploy:x.y.z-gpu-cuda11.4-trt8.4-21.10 -f serving/Dockerfile .
 ```
 
+For example, create an GPU image based on FastDeploy v1.0.0 and ubuntu 20.04，cuda11.2 environment
+```
+# Enter the serving directory and execute the script to compile the FastDeploy and serving backend
+cd serving
+bash scripts/build_fd_cuda_11_2.sh
+
+# Exit to the FastDeploy home directory and create the image
+cd ../
+docker build -t paddlepaddle/fastdeploy:1.0.0-gpu-cuda11.2-trt8.4-20.04 -f serving/Dockerfile_CUDA_11_2 .
+```
+
 ## CPU Image
 
 ```shell

diff --git a/serving/docs/zh_CN/compile.md b/serving/docs/zh_CN/compile.md
@@ -18,6 +18,17 @@ cd ../
 docker build -t paddlepaddle/fastdeploy:x.y.z-gpu-cuda11.4-trt8.4-21.10 -f serving/Dockerfile .
 ```
 
+比如在ubuntu 20.04，cuda11.2环境下制作基于FastDeploy v1.0.0的GPU镜像
+```
+# 进入serving目录执行脚本编译fastdeploy和服务化的backend
+cd serving
+bash scripts/build_fd_cuda_11_2.sh
+
+# 退出到FastDeploy主目录，制作镜像
+cd ../
+docker build -t paddlepaddle/fastdeploy:1.0.0-gpu-cuda11.2-trt8.4-20.04 -f serving/Dockerfile_CUDA_11_2 .
+```
+
 ## 制作CPU镜像
 
 ```

diff --git a/serving/scripts/build_fd_cuda_11_2.sh b/serving/scripts/build_fd_cuda_11_2.sh
@@ -0,0 +1,59 @@
+#!/usr/bin/env bash
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+if [ ! -d "./cmake-3.18.6-Linux-x86_64/" ]; then
+    wget https://github.com/Kitware/CMake/releases/download/v3.18.6/cmake-3.18.6-Linux-x86_64.tar.gz
+    tar -zxvf cmake-3.18.6-Linux-x86_64.tar.gz
+    rm -rf cmake-3.18.6-Linux-x86_64.tar.gz
+fi
+
+if [ ! -d "./TensorRT-8.4.1.5/" ]; then
+    wget https://fastdeploy.bj.bcebos.com/third_libs/TensorRT-8.4.1.5.Linux.x86_64-gnu.cuda-11.6.cudnn8.4.tar.gz
+    tar -zxvf TensorRT-8.4.1.5.Linux.x86_64-gnu.cuda-11.6.cudnn8.4.tar.gz
+    rm -rf TensorRT-8.4.1.5.Linux.x86_64-gnu.cuda-11.6.cudnn8.4.tar.gz
+fi
+
+# build vision、runtime、backend
+docker run -it --rm --name build_fd_libs \
+           -v`pwd`/..:/workspace/fastdeploy \
+           -e "http_proxy=${http_proxy}" \
+           -e "https_proxy=${https_proxy}" \
+           nvidia/cuda:11.2.2-cudnn8-devel-ubuntu20.04 \
+           bash -c \
+           'cd /workspace/fastdeploy/python;
+            rm -rf .setuptools-cmake-build dist build fastdeploy/libs/third_libs;
+            apt-get update;
+            apt-get install -y --no-install-recommends patchelf python3-dev python3-pip rapidjson-dev git;
+            ln -s /usr/bin/python3 /usr/bin/python;
+            export PATH=/workspace/fastdeploy/serving/cmake-3.18.6-Linux-x86_64/bin:$PATH;
+            export WITH_GPU=ON;
+            export ENABLE_TRT_BACKEND=OFF;
+            export TRT_DIRECTORY=/workspace/fastdeploy/serving/TensorRT-8.4.1.5/;
+            export ENABLE_ORT_BACKEND=OFF;
+            export ENABLE_PADDLE_BACKEND=OFF;
+            export ENABLE_OPENVINO_BACKEND=OFF;
+            export ENABLE_VISION=ON;
+            export ENABLE_TEXT=ON;
+            python setup.py build;
+            python setup.py bdist_wheel;
+            cd /workspace/fastdeploy;
+            rm -rf build; mkdir -p build;cd build;
+            cmake .. -DENABLE_TRT_BACKEND=ON -DCMAKE_INSTALL_PREFIX=${PWD}/fastdeploy_install -DWITH_GPU=ON -DTRT_DIRECTORY=/workspace/fastdeploy/serving/TensorRT-8.4.1.5/ -DENABLE_PADDLE_BACKEND=ON -DENABLE_ORT_BACKEND=ON -DENABLE_OPENVINO_BACKEND=ON -DENABLE_VISION=ON -DBUILD_FASTDEPLOY_PYTHON=OFF -DENABLE_PADDLE2ONNX=ON -DENABLE_TEXT=OFF -DLIBRARY_NAME=fastdeploy_runtime;
+            make -j`nproc`;
+            make install;
+            cd /workspace/fastdeploy/serving;
+            rm -rf build; mkdir build; cd build;
+            cmake .. -DFASTDEPLOY_DIR=/workspace/fastdeploy/build/fastdeploy_install -DTRITON_COMMON_REPO_TAG=r21.10 -DTRITON_CORE_REPO_TAG=r21.10 -DTRITON_BACKEND_REPO_TAG=r21.10;
+            make -j`nproc`'