C++ API

This is the API documentation for the NVIDIA TensorRT-RTX library. The NVIDIA TensorRT-RTX C++ API allows developers to import, calibrate, generate, and deploy networks using C++. You can import networks directly from ONNX. You can also create them programmatically by instantiating individual layers and setting parameters and weights directly.

Interface - TensorRT-RTX C++ API

See also

Building and running an engine

End-to-end walkthrough of common C++ API usage patterns.

Working with dynamic shapes

Configuring optimization profiles and dynamic input dimensions.

Using runtime caching

Saving and loading the runtime cache for faster startups.

Operators reference

Full list of supported TensorRT-RTX operators.