cudaErrorInvalidConfiguration with jetson-inference

Hello all,

I am using jetson-inference classes in C++ to load a custom architecture based on detectNet.cpp (in a TX2 4GB), I can perform the pre processing, inference and post processing without any issue, but when I run cudaDetectionOverlay I receive the error cudaErrorInvalidConfiguration. Can this be an issue with the GPU memory ?

Hi,

Here are some details of the error message:
https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html

cudaErrorInvalidConfiguration = 9

This indicates that a kernel launch is requesting resources that can never be satisfied by the current device. Requesting more shared memory per block than the device supports will trigger this error, as will requesting too many threads or blocks. See cudaDevicePropfor more device limitations.

It looks like this is related to the memory resource.
Do you make any modification to the jetson_inference source?

Thanks.

Hi,

Thanks for the documentation. I will check it.

I added a static method in the class detectNet to access cudaDetectionOverlay, I can’t access to this method when compiling my program, I see the message Undefined reference to cudaDetectionOverlay. I am using cmake to build my project, based in the example from https://github.com/dusty-nv/jetson-inference/tree/master/examples/my-recognition.