Build PyTorch with CUDA 11.6

Hi all.
I tried to create PyTorch wheel with cuda 11.6, but faced OOM error.
I was following instructions from here: GitHub - pytorch/pytorch: Tensors and Dynamic neural networks in Python with strong GPU acceleration
After necessary dependencies was installed I tried to build pytorch with command:

python3 bdist_wheel

But soon after compilation started, I get all my PC memory filled and get system error.
How can I solve this issue?

You could limit the number of processes via MAX_JOBS=nb_jobs to avoid running out of memory.

1 Like