Operator Project Build

Overview

After the operator deliverable development is complete, you need to build the operator project to generate a custom operator installation package custom_opp_Linux_Arch.run. The build workflow includes:

• Build the operator information library definition file (.ini) into an operator information library (.json).
• Build the AI CPU operator code implementation files (.h and .cc) into a dynamic library file libcust_aicpu_kernels.so.
• Build the operator plugin implementation files (.h and .cc) into an operator plugin libcust_{tf|caffe|onnx}_parsers.so.
• Build the operator prototype definition files (.h and .cc) into an operator prototype library libcust_op_proto.so.

The AI CPU operator project of an earlier version is incompatible with the latest version of the Ascend-CANN-Toolkit. You need to modify the CMakeLists.txt file and rebuild the operator project. Perform the following steps to solve the problem:

1. Modify the CMakeLists.txt file in the /cpukernel directory of the operator project.

   ...
   set(AICPU_SOC_VERSION $ENV{AICPU_SOC_VERSION}) 
   message(STATUS "AICPU_SOC_VERSION=${AICPU_SOC_VERSION}")
   
   # Add the following code.
   if(EXISTS "${ASCEND_AICPU_PATH}/opp/built-in/op_impl/aicpu/aicpu_kernel/lib/${AICPU_SOC_VERSION}/libascend_protobuf.a")
       target_link_options(${AICPU_KERNEL_TARGET} PRIVATE
           -s 
           -Wl,-Bsymbolic
           -Wl,--exclude-libs=libascend_protobuf.a
       ) 
       target_link_libraries(${AICPU_KERNEL_TARGET} PRIVATE
           -Wl,--whole-archive
           ${ASCEND_AICPU_PATH}/opp/built-in/op_impl/aicpu/aicpu_kernel/lib/${AICPU_SOC_VERSION}/libascend_protobuf.a
           -Wl,--no-whole-archive
       ) 
   endif()
   
   if(EXISTS "${ASCEND_AICPU_PATH}/opp/built-in/op_impl/aicpu/aicpu_kernel/lib/${AICPU_SOC_VERSION}/libcpu_kernels_context.a")
   ...

2. Rebuild the operator project by referring to this section.

Build Procedure

1. (Optional) Configure the path of the third-party library referenced by the AI CPU operator by modifying the cpukernel/CMakeLists.txt file in the operator project directory.
   • include_directories: Add the directories of the header files to be included.

     The following is an example:

     include_directories(
     directoryPath1
     directoryPath2
     )

   • link_directories: Add the directories of the library files to be linked with.

     The following is an example:

     link_directories(
     directoryPath3 
     directoryPath4
     )

   • link_libraries: Add library files for operator implementation.

     The following is an example:

     link_libraries( 
     libName1 
     libName2)

   

   For details about the CMake parameters, visit https://cmake.org/cmake/help/latest/guide/tutorial/index.html and query the required version.

2. On the MindStudio IDE project page, select the operator project and choose Build > Edit Build Configuration... from the top menu bar.
3. Go to the Build Configuration dialog box. Click to add configurations. By default, the Release(default) build type is added. For details, see Table 1.
   Figure 2 Build configurations dialog box
   

   Table 1 Description of build configuration parameters

   Parameter	Description
   Name	Custom configuration name. Defaults to Release.
   Build type	Configuration type, which is optional. Defaults to Release.
   Toolchain	Toolchain configurator, which preconfigures a custom toolchain with the same architecture as the installed CANN package and supports local build. You can click Manage toolchains... to customize the toolchain. For details, see Toolchains.
   CMake options	CMake option. The default value is -DCMAKE_BUILD_TYPE=Release.
   Build directory	Relative path of the build directory, which is relative to the project directory.
   Build options	Build acceleration option.
   Environment variables	Environment variable configuration, which can be set before project build. You can manually configure the environment variables or click to configure them in the dialog box displayed.
   Environment from path	Enter a path or click on the right to select the environment variable configuration file. The configuration file is filled in with one or more lines of <variable name>=<variable value>, for example: APATH=/usr/local/xxx X_PATH=/xxx/xxx

   

   You are advised to configure the AICPU_KERNEL_TARGET environment variable and add the software version number as a suffix to the name of the dynamic library file to avoid library file naming conflicts due to AI CPU upgrade.

   If this environment variable is not set, the default value cust_aicpu_kernels is used.

   During build, add the environment variable to the OS.

   export ASCEND_AICPU_PATH=Ascend-CANN-Toolkit installation directory/ascend-toolkit/latest
   export AICPU_KERNEL_TARGET=cust_aicpu_kernels_3.3.0 (recommended)

4. Click or choose Build > Build Ascend Operator Project to start the build.
5. View the build result in the lower part of MindStudio. The custom operator installation package custom_opp_Linux_Arch.run is generated in the cmake-build directory of the operator project.

   The value of Arch is obtained based on the CANN package and toolchain information.