Background
After the hackathon, I started to investigate some of the platform/architecture issues that many of the participants encountered. Since I am running primarily x86_64 architectures, I seemed to have been able to run AMPL without any issues. However, after working with @andriialbatov, who is running an arm64, we noticed that there are certain python packages which don't exist for this architecture. This makes some sense, since no arm64 machines are equipped with nvidia cards
Problematic Dependencies
- jax [cpu]
- jaxlib [cpu]
- torch [cpu]
- nvidia-cublas-cu11 [gpu]
References
https://stackoverflow.com/questions/63423463/using-pytorch-cuda-on-macbook-pro
Background
After the hackathon, I started to investigate some of the platform/architecture issues that many of the participants encountered. Since I am running primarily
x86_64architectures, I seemed to have been able to run AMPL without any issues. However, after working with @andriialbatov, who is running anarm64, we noticed that there are certain python packages which don't exist for this architecture. This makes some sense, since noarm64machines are equipped with nvidia cardsProblematic Dependencies
References
https://stackoverflow.com/questions/63423463/using-pytorch-cuda-on-macbook-pro