Build with onnxruntime

This package includes a wrapper for onnxruntime based on cython. The standard one relies on pybind11. For that purpose, it includes the onnxruntime binaries released on github (see onnxruntime releases).

cmake

The first step is to load the extension FindOrt.cmake with find_package(Ort REQUIRED). This file exposes two functions. The first one ort_add_dependency(name folder_copy) copies the binaries into folder folder_copy and links target name with onnxruntime.

The second function ort_add_custom_op(name folder) creates a library with several custom kernels for onnxruntime and links it with onnxruntime. name is the project name, folder its location.

ort_add_custom_op(
    ortops_tutorial_cpu                                             # name
    ../onnx_extended/ortops/tutorial/cpu                            # folder
    ../onnx_extended/ortops/tutorial/cpu/my_kernel.cc               # source file
    ../onnx_extended/ortops/tutorial/cpu/my_kernel_attr.cc          # source file
    ../onnx_extended/ortops/tutorial/cpu/ort_tutorial_cpu_lib.cc)   # source file

Every new kernel can be added by adding new source file. A line must be added in file ort_tutorial_cpu_lib.cc to register the kernel. That file also defines the domain the kernel belongs to.

This function is subject to change. It creates a file _setup_ext.txt to indicate which file to copy from the build directory to the package directory. This file is loaded by setup.py after cmake is done with the compilation.