Error encountered when compiling cuML on Jestson AGX Orin

[WRangers]

Hi, there.

Thank you firstly for your cuML containers. However, when I tried to compile a newer version, e.g. cuML 24.04, based on the container nvcr.io/nvidia/l4t-cuda:11.4.19-devel, I encountered an error when compiling the raft, which should not be an error --- a variable was declared but never used. I also tried to modify the CMakeLists.txt, but failed.

Later, I gave up compiling from source code, installing cuML 24.04 via conda according to https://docs.rapids.ai/install via
conda create -n rapids-24.04 -c rapidsai -c conda-forge -c nvidia cuml=24.04 python=3.9 'cuda-version==11.4'. Though the installation was successful, the excution of CPP DBSCAN example failed (verbose output is listed in the end).

My physical device information:

• CUDA Arch BIN: 8.7
• L4T 35.4.1
• Jetpack: 5.1.2
• Based container: nvcr.io/nvidia/l4t-cuda:11.4.19-devel from https://catalog.ngc.nvidia.com/orgs/nvidia/containers/l4t-cuda/tags

I wonder how you built cuML successfully before. Could you please give me some advice? Thank you very much!

error.txt