Build TensorRT statically

I am working on statically building TensorRT on my Windows system. The goal is to reduce the size of my program by eliminating the need for dynamic libraries (DLLs) and ensuring that only the necessary parts of the libraries are included in the final program.

For this, I have been attempting to build TensorRT from source in static mode. I am using CMake to generate build files, and while the solution files are generated successfully, when I open the solution in Visual Studio, I do not see any targets for nvinfer or nvinfer_builder_resource, which seem to be crucial libraries for TensorRT.

Here’s a summary of the issues I am facing:

Missing Targets: The targets for nvinfer and nvinfer_builder_resource are not present in the generated Visual Studio solution. I am unable to locate where these are configured in the CMakeLists.txt file.

Feasibility of Static Build: I am also wondering if it is possible to build TensorRT statically on Windows. If so, could you provide any official documentation or guidance on how to achieve this?

Below is an image of the generated targets where nvinfer and nvinfer_builder_resource are missing.

Thank you in advance for your help and guidance.

Start by reviewing the CMakeLists.txt configuration that generates the Windows Visual Studio solution and the options controlling static builds. Check whether nvinfer and nvinfer_builder_resource are defined or intentionally omitted, then document whether static TensorRT builds on Windows are supported and what configuration is required.