Does Pytorch >1.12 support aarch64?

I have aarch64 cpu, and cuda11.4.
so I use

conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch

command but i got the following error.

PackagesNotFoundError: The following packages are not available from current channels:

  - torchaudio==0.12.1
  - cudatoolkit=11.3

Current channels:


is this error related to aarch64 architecture?

and when I type nvidia-smi, It shows

NVIDIA-SMI 510.47.03 Driver Version: 510.47.03 CUDA Version: 11.6
and 2 of NVIDIA A100 GPU

but for nvcc --version, I got

Cuda compilation tools, release 11.4, V11.4.100
Build cuda_11.4.r11.4/compiler.30188945_0

does this inconsistency of CUDA version cause any error like the above?
Thank you.

I don’t think ARM binary builds are provided (with the exception for Mac) and I guess you are using an ARM server CPU?
If so, you might need to build PyTorch from source or use the NGC container.

Thank you.
yes, I’m using ARM server CPU. then can I build PyTorch from source on Conda environment?

Yes, you can build PyTorch from source, too, if needed.