onnx_extended.__init__.py#

check_installation#

onnx_extended.check_installation(ortops: bool = False, ortcy: bool = False, val: bool = False, verbose: bool = False)[source]#

Quickly checks the installation works.

Parameters:

  • ortops – checks that custom ops on CPU are working

  • ortcy – checks that OrtSession is working (cython bindings of onnxruntime)

  • val – checks that a couple of functions in submodule validation are working

  • verbose – prints out which verifications is being processed

compiled_with_cuda#

onnx_extended.compiled_with_cuda() → bool[source]#

Checks it was compiled with CUDA.

cuda_version#

onnx_extended.cuda_version() → str[source]#

Tells which version of CUDA was used to build the CUDA extensions.

cuda_version_int#

onnx_extended.cuda_version_int() → tuple[source]#

Tells which version of CUDA was used to build the CUDA extensions. It returns (0, 0) if CUDA is not present.

get_cxx_flags#

onnx_extended.get_cxx_flags() → str[source]#

Returns CXX_FLAGS.

get_stdcpp#

onnx_extended.get_stdcpp() → int[source]#

Returns CMAKE_CXX_STANDARD.

has_cuda#

onnx_extended.has_cuda() → bool[source]#

Tells if cuda is available.