meet Segmentation fault when "Building CUDA object CMakeFiles/main.dir/main.cu.o"

[xmaww]

Hello! Thank you for the open source code!
I try to configure and build the project on my own computer with ubuntu 20.04 and one Segmentation fault continues occuring when I execute
cmake --build build -- -j1 # build
as shown in fig1:

And the corresponding build.make code is shown in fig2:

Computer configuration:

1. 64G of RAM.
2. Graphics card RTX 3080 (10G version).
3. cuda version 12.0.
4. nvcc -V and nvidia-smi commands work.
5. Tested cuda in .cu scripts.
   Looking forward to your kind reply, thanks! @Ailon-Island

[Ailon-Island]

Hello,
it's nice to know that you are trying TacIPC!
Similar problems do not occur in our environments. Therefore, rather than a solution, here are just some suggestions that might help.

1. Since the command cmake --build build -- -j1 has an empty prompt indicator --, (or an end-of-command indicator), you might double check if the given line is really what you executed and does what you mean.
2. Actually, to build the project, we recommend cmake --build build --parallel [N_threads], where [N_threads] can be any proper number of threads. Conservatively 8, for example. Segfaults sometimes occur due to too many threads.
3. All other configurations seem fine. But does 5. in your computer configuration mean that you've tested building other CUDA projects with cmake? If so, there shouldn't be much problem with your environment configuration. Otherwise, maybe you could try to build some light and simple CUDA projects with cmake if suggestions above does not work.
4. (update) Maybe you could try newer cmake. On ubuntu, you can install with snap install cmake --classic and use the installed binary /snap/bin/cmake instead of that installed with apt.