How can I plant the libtorch on NVIDIA TX2?

C++
1

I planted the project on NVIDIA TX2, cmake was completely fine, but error came when makefile completed 100%, the information is that"/path/to/libtorch/lib/libtorch.so: error adding symbols: File in wrong format".
i search the error but little useful information could be found, How can i plant the libtorch projects on NVIDIA TX2?

2

diud you download libtorch binaries from the PyTorch website, and are you using them for TX2? If that’s the case, this will not work, as libtorch binaries listed on the website are for x64 architecture (but TX2 is ARM architecture).

To build for TX2, you have to build libtorch from source. See if this thread helps: TX2 (ARM64) Install for Python 3 - Cont from nVidia Forum

1 Like
3

thanks for suggestion, i am trying to build libtorch from source:grin::grin: