基于onnxruntime的C++版本TensorRt源码编译

基于onnxruntime的C++版本TensorRt源码编译

提示：基于onnxruntime的C++版本TensorRt源码编译

---

---

一、源码地址

onnxruntime源码github地址.
https://github.com/microsoft/onnxruntime

Tensorrt7.2.2.3
https://developer.nvidia.com/nvidia-tensorrt-7x-download

---

二、步骤

1.基础环境搭建

• Ubuntu18.04
• cmake version 3.20.3
• python3.6.9(一般系统就自带，可以安装3.8的)
  ps:onnxruntime的tensorRt的编译需要下载tensorRt7.2.2.3,支持的python版本:2.7,3.4-3.8,目前暂不支持3.9以上
• Tensorrt7.2.2.3
• onnxruntime-tensorrt官网编译指导说明
• nvidia驱动
  CUDA Toolkit 11.0 Download
  Download cuDNN v8.0.5 (November 9th, 2020), for CUDA 11.0
  

---

2. 安装配置

# cuda安装包直接执行
./cuda_11.0.2_450.51.05_linux.run

# cuDNN安装
tar -zxvf cudnn-11.0-linux-x64-v8.0.5.39.tgz
sudo cp -rf cuda/include/cudnn* /usr/local/cuda/include/
sudo cp -rf cuda/lib64/libcudnn* /usr/local/cuda/lib64/
sudo chmod a+r /usr/local/cuda/include/cudnn*
sudo chmod a+r /usr/local/cuda/lib64/libcudnn*

dpkg -i libcudnn8_8.0.5.39-1+cuda11.0_amd64.deb
dpkg -i libcudnn8-dev_8.0.5.39-1+cuda11.0_amd64.deb
dpkg -i libcudnn8-samples_8.0.5.39-1+cuda11.0_amd64.deb

# Tensorrt7.2.2.3安装
tar -zxvf TensorRT-7.2.2.3.Ubuntu-18.04.x86_64-gnu.cuda-11.0.cudnn8.0.tar.gz -C /home/
# 解压后,按照官网[4.5 Tar File Installation安装说明]

Tensorrt Tar File Installation

---

3.设置环境变量

# vim /etc/profile
export PATH=$PATH:/usr/local/cuda-11.0/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.0/lib64
export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/cuda-11.0/lib64

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/TensorRT-7.2.2.3/lib
export PATH=$PATH:/home/TensorRT-7.2.2.3

# 保存退出
source /etc/profile

---

三、源码编译

./build.sh --build_shared_lib --skip_tests --config Release --use_cuda --cudnn_home /usr/local/cuda/ --cuda_home /usr/local/cuda  --use_tensorrt --tensorrt_home /home/TensorRT-7.2.2.3/

编译过程展示


onnxruntime_test

下面就是编译完成的库,在onnxruntime/build/Linux/Release中

---

四、测试

暂时不展示，后续会放在个人github上