Fail to build C++ SDK when use the cmake command

[yeungkong]

Hi developers of mmdeploy. My environment is Win10, Cuda11.3, VS2017, cudnn8.2.0, Python3.7, Pytorch1.11.0, mmcv-full1.4.8, mmdet2.23.0. I can convert my model to tensorrt successfully in Win 10. But I encountered errors when I try to use the cmake command to build the SDK. The following is the command.

mkdir build
cd build
cmake ..\cpp -A x64 -T v141 
-DMMDEPLOY_BUILD_SDK=ON 
-DMMDEPLOY_BUILD_SDK_CXX_API=ON 
-DMMDEPLOY_BUILD_TEST=ON 
-DMMDEPLOY_TARGET_DEVICES="cuda" 
-DMMDEPLOY_TARGET_BACKENDS="trt" 
-DMMDEPLOY_CODEBASES="all" 
-DONNXRUNTIME_DIR="D:\...\onnxruntime" 
-DOpenCV_DIR="D:\opencv2\opencv\build\x64\vc15\lib" 
-DMMDeploy_DIR="C:\Users\Administrator\source\repos\C++_And_Python\mmdeploy-0.7.0-windows-amd64-onnxruntime1.8.1\sdk\lib\cmake\MMDeploy" 
-DTENSORRT_DIR="C:\Program Files\NVIDIA GPU Computing Toolkit\tensorrt\TensorRT-8.4.3.1" 
-DCUDNN_DIR="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.3"

The following is the error message.

-- Building for: Visual Studio 15 2017
-- The C compiler identification is MSVC 19.16.27048.0
-- The CXX compiler identification is MSVC 19.16.27048.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio/2017/Community/VC/Tools/MSVC/14.16.27023/bin/Hostx86/x64/cl.exe - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2017/Community/VC/Tools/MSVC/14.16.27023/bin/Hostx86/x64/cl.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Found OpenCV: D:/opencv2/opencv/build (found version "3.4.14")
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Failed
-- Looking for pthread_create in pthreads
-- Looking for pthread_create in pthreads - not found
-- Looking for pthread_create in pthread
-- Looking for pthread_create in pthread - not found
-- Found Threads: TRUE
CMake Error at CMakeLists.txt:6 (target_link_libraries):
  Cannot specify link libraries for target "PRIVATE" which is not built by
  this project.


-- Configuring incomplete, errors occurred!
See also "C:/Users/Administrator/source/repos/C++_And_Python/mmdeploy-0.7.0-windows-amd64-onnxruntime1.8.1/sdk/example/build/CMakeFiles/CMakeOutput.log".
See also "C:/Users/Administrator/source/repos/C++_And_Python/mmdeploy-0.7.0-windows-amd64-onnxruntime1.8.1/sdk/example/build/CMakeFiles/CMakeError.log".

In order to convert model successfully, I strictly follow the instructions to set up the mmdeploy environment and it works. Is something wrong with my execuation? Please help me with the error. Thank you very much.

[irexyc]

Please use mmdeploy-0.7.0-windows-amd64-cuda11.1-tensorrt8.2.3.0.zip instead of mmdeploy-0.7.0-windows-amd64-onnxruntime1.8.1

It seems like that you are building examples. Your cmake command is not right. To build examples, there is no need to pass DMMDEPLOY_BUILD_SDK, DMMDEPLOY_BUILD_SDK_CXX_API, DMMDEPLOY_BUILD_TEST, etc.

For how to build examples, you can refer to this doc

[yeungkong]

@irexyc I am trying to build C++ SDK environment so that I can call my model by C++. Is that what you mean examples?

[irexyc]

@yeungkong Your cmake command not only pass the MMDeploy_DIR option，but also pass MMDEPLOY_BUILD_SDK option. It seems that there is no document tells that. Which document do you follow?

If you want to build mmdeploy and use the built mmdeploy to build mmdeploy-examples，please follow this doc

If you want to use the prebuilt package to build exampels, please follow this doc

According to your cmake log, your working dir is C:/Users/Administrator/source/repos/C++_And_Python/mmdeploy-0.7.0-windows-amd64-onnxruntime1.8.1/sdk/example. It seems that you are using the prebuilt package, is that right? If so, you use the wrong cmake command and you should use mmdeploy-0.7.0-windows-amd64-cuda11.1-tensorrt8.2.3.0 instead.

你是不是在用预编译包，然后想编译预编译包中的example？如果是的话，你用错预编译包了，如果要支持trt，你应该用mmdeploy-0.7.0-windows-amd64-cuda11.1-tensorrt8.2.3.0这个包，然后编译命令也不对，似乎没有文档是你那种用法，具体的可以参考上面列出来的两个文档。

[lvhan028]

Hi, @yeungkong it is still an issue for you?