Skip to content
Branch: master
Find file Copy path
Find file Copy path
6 contributors

Users who have contributed to this file

@RyanUnderhill @pranavsharma @jignparm @torkeldanielsson @rgesteve @askhade
32 lines (24 sloc) 1.84 KB



  • Creating an InferenceSession from an on-disk model file and a set of SessionOptions.
  • Registering customized loggers.
  • Registering customized allocators.
  • Registering predefined providers and set the priority order. ONNXRuntime has a set of predefined execution providers, like CUDA, MKLDNN. User can register providers to their InferenceSession. The order of registration indicates the preference order as well.
  • Running a model with inputs. These inputs must be in CPU memory, not GPU. If the model has multiple outputs, user can specify which outputs they want.
  • Converting an in-memory ONNX Tensor encoded in protobuf format to a pointer that can be used as model input.
  • Setting the thread pool size for each session.
  • Setting graph optimization level for each session.
  • Dynamically loading custom ops. Instructions
  • Ability to load a model from a byte array. See OrtCreateSessionFromArray in onnxruntime_c_api.h.

Usage Overview

  1. Include onnxruntime_c_api.h.
  2. Call OrtCreateEnv
  3. Create Session: OrtCreateSession(env, model_uri, nullptr,...)
    • Optionally add more execution providers (e.g. for CUDA use OrtSessionOptionsAppendExecutionProvider_CUDA)
  4. Create Tensor
    1. OrtCreateAllocatorInfo
    2. OrtCreateTensorWithDataAsOrtValue
  5. OrtRun

Sample code

The example below shows a sample run using the SqueezeNet model from ONNX model zoo, including dynamically reading model inputs, outputs, shape and type information, as well as running a sample vector and fetching the resulting class probabilities for inspection.

You can’t perform that action at this time.