How much GPU ram is available when I train model in M1 Macbook? #37
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
How much GPU ram is available when I train model in M1 Macbook?
Hi @Dfaker-HK!
I'm not familiar enough with latest Mac's hardware, so can't give you an expert advise, but I find this discussion quite helpful as it gives couple good links for further reading.
As far as I understand, in the new Mac's CPU and GPU have shared RAM, so I would expect that GPU can use all the free RAM that you have on your laptop given that it runs the OS and all the processes.
May I ask how you did get it running on M1? I am struggling with the python libraries.
@bufo24 I didn't get chance to test it on M1. Could you please open a separate issue and provide details of what is your Python version and what's the error output you're getting?
I am not getting the error actually
I just plan to buy the new m1 macbook and use it to train the model 7x24 as my 3090 is too loud that I can’t let it run when I am sleeping.
And I want to know what model can I run on it.
The issue is tagged as "stale issue" due to no activity in 30 days. If there is no activity for 15 more days, the issue will be closed.
I am trying to run on my MBP M1 Pro and the lack of tensorflow version for the M1 is killing me! Did anyone solve that? All I get is:
Edit:
My workaround was: installing the tensorflow-macos and tensorflow-metal, and for PyQt5, I installed from brew (
brew install pyqt@5), copied the files inside site-packages from the Homebrew/Caskroom folders and paste it on the env folders of DeepfacelabMacOS. Now it works.@mateuspestana what Python version do you have? Currently only Python up to
3.9(included) is supported. The3.10is not supported yet. If you have3.10, you can try using pyenv to switch between3.9and3.10. When you switch to3.9, make sure that bothpythonandvirtualenvpoint at the same version (check withpython -Vandvirtualenv --version).I'm on 3.9!
@mateuspestana Hmm, interesting. Unfortunately, I don't have M1 to troubleshoot it myself. Do you mind sharing full console output just in case? Would be great if you also remove the
.dfl/dir to make sure you're starting from scratch.What I do after that so it works: I install PyQt5 from brew and put the files inside the env (and remove the line from the requirements and run the rest), so when it runs, it gets "sees" version and can import.
The issue is tagged as "stale issue" due to no activity in 30 days. If there is no activity for 15 more days, the issue will be closed.
Closing this issue due to no activity in last 45 days