Including custom code to NVIDIA Jetson Platform

Hi,

I'm trying to add custom C code to NVIDIA Jetson Support package, but I'm facing issues when compiling the code.

I have followed the help on integrating custom C code for Raspberry Pi Driver Device and the code is generated. However, when NVCC tries to compile it, it looks for standard libraries in the Jetson Platform but can't find it.

The error I'm getting is:

fatal error: algorithm: No such file or directory
66 | #include <algorithm>
|          ^~~~~~~~~~~
% 

The inlcude is actually inside the cpp file for a sensor SDK.

So the workflow is:

Simulink Block points to my cpp > my cpp includes the Sensor SDK > Sensor SDK includes cuda.h and standard libraries, such as algorithm.

Another odd behavior is when I just add the path with the buildInfo.addIncludePaths, the build doesn't find the libraries. I have to write the full path on my cpp file. The it finds the library. But I can't do it for all the std libraries.

How to solve this?

Best,

Answer 1

[David Fink]

This could be a C++ compiler installation issue (std library headers aren't automatically on the include path), a build configuration issue (if somehow NVCC is compiling with a C compiler instead of a C++ compiler), or something else.

Please reach out to MathWorks Technical Support with more information (including full reproduction steps) for further assistance.