int-brain-lab / iblenv

Unified environment and Issue tracker for all IBL
MIT License
10 stars 10 forks source link

[Bug report] - Slow performance of ONE on M1 mac native python (not urgent) #295

Open berkgercek opened 2 years ago

berkgercek commented 2 years ago

I am running ONE on an M1 Macbook, using python with binaries native to the darwin OSX ARM architecture. When I try to run e.g. downloads using ONE (for example, downloading the Allen Atlas after importing ibllib.atlas.AllenAtlas), the processing is extremely slow to the point of unusability. The example download has taken >10m and is still not complete. I suspect it might have something to with the threading used by ONE, but this is a shot in the dark.

I know this is a very niche issue, so I'm not really looking for a quick fix or anything but I think it's something the devs should be aware of.

My environment spec can be found in the iblenv_osx-arm63.yaml file, installed via mambaforge for ARM macs.

oliche commented 2 years ago

Hey Berk, I have a M1 mac here, the smallest possible, and don't have this issue. You're saying that just importing the atlas shows the issue ?

I use conda as well, but the main yaml file !

Let me know if you want me to dig further config files on my end

berkgercek commented 2 years ago

Hmmm, very strange. Are you running the terminal or your editor via Rosetta? Because for me the main yaml file wouldn't work at all because there were no arm compiled versions available of e.g. mayavi and pyqt. I made that modified env so I could get native python to work