Is this page helpful?

Installing TensorRT-RTX#

TensorRT-RTX can be installed from an SDK zip file on Windows, a tarball on Linux, or via PyPI for Python workflows. Before proceeding, ensure you have met all Prerequisites.

  1. Download the Windows zip package that matches your CUDA Toolkit version and CPU architecture.

  2. Decompress the Windows zip package to extract the following:

    • Core library and ONNX parser DLLs and import libraries

    • Development headers

    • Source code samples

    • Documentation

    • Python bindings

    • The tensorrt_rtx executable for building engines and running inference from the command line

    • Licensing information and open-source acknowledgments

    Note

    The DLLs are signed and verified with the signtool utility. Add the package lib and bin directories to your PATH so Windows can find the DLLs and tensorrt_rtx.exe:

    $trtRoot = "C:\path\to\TensorRT-RTX-1.6.0.0"  # Replace with your extract path
$env:PATH = "$trtRoot\lib;$trtRoot\bin;$env:PATH"

  3. Confirm the CLI resolves:

    tensorrt_rtx.exe --help

    If this fails with a “not recognized” or missing-DLL error, the lib and bin directories are not on your PATH. Refer to If a Command Fails.

  4. Optionally, install the Python bindings.

    $version = "1.6.0.0" # Replace with newest version
$arch = "amd64" # Replace with your architecture
$pyversion = "311" # For Python 3.11. Replace with your
                   # Python version for other versions.
$wheel = Join-Path "TensorRT-RTX-$version" "tensorrt_rtx-$version-cp$pyversion-none-win_$arch.whl"
python3 -m pip install $wheel

TensorRT-RTX can be installed from a tarball package on Linux. For supported distributions and CUDA requirements, refer to the Prerequisites page.

  1. Download the tarball that matches your operating system version and CPU architecture.

  2. Unzip the tarball, optionally adding the path of the executable to your PATH variable and the path of the library to your LD_LIBRARY_PATH variable.

    version="1.6.0.0" # Replace with newest version
arch="x86_64"      # Replace with your architecture
cuda="12.9"        # Replace with your CUDA version
tarfile="TensorRT-RTX-${version}.Linux.${arch}-gnu-${cuda}.tar.gz"
tar -xzf $tarfile
export PATH=$PATH:$PWD/TensorRT-RTX-${version}/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PWD/TensorRT-RTX-${version}/lib

  3. Confirm the CLI resolves:

    tensorrt_rtx --help

    If this fails with command not found or a missing shared-library error, PATH or LD_LIBRARY_PATH is not set for the current shell. Refer to If a Command Fails.

  4. Optionally, install the Python bindings.

    pyversion="311" # Assuming Python 3.11, else replace with your
                # Python version
wheel="tensorrt_rtx-${version}-cp${pyversion}-none-linux_${arch}.whl"
python3 -m pip install TensorRT-RTX-${version}/python/${wheel}

To use TensorRT-RTX from Python, you can install the tensorrt-rtx package directly from PyPI. This provides the Python bindings and the TensorRT-RTX shared library.

On Linux images that ship an older system pip (for example Ubuntu 22.04 with pip 22.x), upgrade pip first. Older pip versions can reject the TensorRT-RTX source distribution with an inconsistent package-metadata error.

python3 -m pip install --upgrade pip
python3 -m pip install tensorrt-rtx

Confirm the import:

python3 -c "import tensorrt_rtx; print(tensorrt_rtx.__version__)"

Note

  • If you installed TensorRT-RTX from the Windows SDK or Linux tarball, you do not need to install the PyPI package separately; the local wheel included in those distributions provides the same Python bindings.

  • The PyPI package does not include the tensorrt_rtx CLI. For the Quick Start Guide walkthrough, use the Windows SDK zip or Linux tarball tab instead.

  • Python inference examples that use pycuda also require a CUDA Toolkit install so pycuda can build or find CUDA headers. Refer to Prerequisites.

Next Steps#

Windows SDK or Linux tarball

After installation, confirm the CLI responds (tensorrt_rtx --help on Linux or tensorrt_rtx.exe --help on Windows), then:

PyPI (Python)

After pip install, confirm python3 -c "import tensorrt_rtx" succeeds, then:

All install methods

On this page